{"schema":"https://paulantoinetual.fr/en/agents#catalogue","language":"en-GB","title":"Paul-Antoine Tual’s blog: AI insights for business leaders","description":"Catalogue of the blog’s articles: AI strategy, agents, digital sovereignty, costs. Each article is available as a web page, as Markdown and as an A2UI surface.","author":{"name":"Paul-Antoine Tual","url":"https://paulantoinetual.fr"},"counts":{"articles":52,"words":96439,"sources":581,"figures":50},"endpoints":{"html":"https://paulantoinetual.fr/en/blog","rss":"https://paulantoinetual.fr/en/rss.xml","jsonFeed":"https://paulantoinetual.fr/en/feed.json","a2ui":"https://paulantoinetual.fr/en/blog/a2ui.json","agents":"https://paulantoinetual.fr/en/agents","llms":"https://paulantoinetual.fr/llms.txt","llmsFull":"https://paulantoinetual.fr/en/llms-full.txt"},"categories":[{"key":"methode","label":"Junyr Method™","count":14,"url":"https://paulantoinetual.fr/en/blog/category/junyr-method"},{"key":"agents","label":"AI Agents","count":13,"url":"https://paulantoinetual.fr/en/blog/category/ai-agents"},{"key":"transition","label":"Transition Management","count":9,"url":"https://paulantoinetual.fr/en/blog/category/transition-management"},{"key":"marches","label":"Markets & AI","count":7,"url":"https://paulantoinetual.fr/en/blog/category/markets-and-ai"},{"key":"efficacite","label":"Personal Effectiveness","count":4,"url":"https://paulantoinetual.fr/en/blog/category/personal-effectiveness"},{"key":"cyber","label":"Cybersecurity","count":2},{"key":"ianative","label":"2027 AI-native Company Insights","count":1},{"key":"exemple","label":"Illustrative example","count":1},{"key":"methodologie","label":"Methodology","count":1}],"topics":[{"key":"agents","label":"Agents and orchestration"},{"key":"couts","label":"Costs and ROI"},{"key":"souverainete","label":"Sovereignty and security"},{"key":"marches","label":"Markets and strategy"},{"key":"methode","label":"Method and change"},{"key":"prompts","label":"Prompts and writing"},{"key":"productivite","label":"Personal effectiveness"},{"key":"outils","label":"Tools and interfaces"}],"articles":[{"slug":"ia-frontiere-gaz-ia-bureau-frugale","title":"The frontier runs on gas. Most office AI does not have to.","description":"The FT reports Anthropic agreed to pay $3.4bn for 1.7 GW of gas generators. Our bet: much office AI can run on a small local model. Evidence for and against.","category":"AI Agents","categoryKey":"agents","published":"2026-11-10","tags":["frugal AI","local AI","open-weight models","AI energy","data centres","gas generators","solar power","AI harness","model routing","Claude Fable 5.2","Claude Sonnet 5.5","SME","Qwen3.8","Apple Mac Studio","Artificial Analysis"],"words":2536,"readingMinutes":12,"sources":16,"figures":3,"topics":{"agents":5,"couts":3,"marches":3},"faq":[{"question":"Did Anthropic buy gas turbines?","answer":"The available reports describe generators and no announcement. The Financial Times reported on 8 November 2026, as relayed by The Next Web, that Anthropic agreed to pay $3.4bn for 1.7 GW of gas-fired generators made by Caterpillar, by taking over orders placed by OnePWR Solutions."},{"question":"Can a small open model replace Claude Sonnet 5.5 for office work?","answer":"It has not been shown. On the Artificial Analysis Intelligence Index v5 of 27 October 2026, Sonnet 5.5 scores 55 and the best released open-weight models score 32. A small open model can win on a narrow task, which is why Junyr treats office work as a hypothesis to test task by task."},{"question":"How much solar power does a 100 W AI machine need?","answer":"As illustrative arithmetic, 100 W running all year uses 876 kWh. With PVGIS data, an annual balance needs about 0.78 kWp of panels in Paris and 0.54 kWp in Marseille, and about 1.8 kWp to cover a Paris December, plus a battery for the nights."},{"question":"What should an SME measure before running AI on its own machines?","answer":"Three things, in order: which share of its real requests is routine, how a local model and a cloud model compare on those requests inside the same harness, and how many watts and euros the local machine actually uses."},{"question":"What is model routing, and how does it apply to office AI?","answer":"Model routing sends each request to the smallest model that can handle it and escalates only when needed. Our design has three tiers, a small local model (planned), a local 27-billion-parameter model and a cloud frontier model, with a harness choosing between them; no share of traffic per tier has been measured."},{"question":"Is running AI locally cheaper than paying per token?","answer":"We cannot say yet. One refurbished Mac Studio M4 Max costs us about €2,000 excluding VAT, about €56 a month over 36 months before electricity, while cloud list prices run from $10 to $40 per million output tokens; without our token volume and a wall-meter reading, no price per token is publishable."}],"url":"https://paulantoinetual.fr/en/blog/ia-frontiere-gaz-ia-bureau-frugale","markdownUrl":"https://paulantoinetual.fr/en/blog/ia-frontiere-gaz-ia-bureau-frugale.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/ia-frontiere-gaz-ia-bureau-frugale.a2ui.json","image":"https://paulantoinetual.fr/img/blog/ia-frontiere-gaz-og-en.png"},{"slug":"negation-majuscules-questions-prompts-ce-que-la-preuve-soutient","title":"“Don't do that”, “VERY IMPORTANT”, “Can you…?”: what the evidence says about negation, capitals and questions in prompts","description":"Few studies test two phrasings of the same instruction inside an actual agent prompt. We reviewed the 2026 vendor guides, the papers and the field reports on negation, capital letters, repetition and the question form, and kept only what they state: where the evidence is strong, where it is thin, and which circulating figures we discarded.","category":"AI Agents","categoryKey":"agents","published":"2026-11-06","tags":["prompt engineering","negation","capital letters","prompt repetition","clarifying questions","AI agents","instruction following","interpretability","Claude","GPT-6 Astra"],"words":2610,"readingMinutes":12,"sources":23,"figures":3,"topics":{"agents":6,"prompts":10,"outils":3},"faq":[{"question":"Is it a mistake to tell an AI model what not to do?","answer":"Not always. Short avoidance rules such as 'no commas' held at or near 100% under task load in one study, but broad prohibitions are over-applied, and naming a forbidden word can prime it. Anthropic advises saying what to do instead."},{"question":"Do capital letters and words like CRITICAL make a model follow a rule better?","answer":"On recent Claude models, Anthropic says they cause over-triggering and advises plain wording. A 13-model study found capitals raised accuracy on factual questions by up to 8.95 points, but no study isolates capitals on behavioural rules."},{"question":"Does repeating an instruction help?","answer":"Repeating the whole prompt helped models that do not reason, in a study run on 2025 models. A short reminder of a format constraint at the end of the prompt restored 90 to 100% compliance in many settings in another. OpenAI advises stating each rule once."},{"question":"Does 'Can you…?' make an AI agent ask more questions?","answer":"No study shows it. 'Can you suggest…' can get advice where an imperative gets the action, but the interruptions we found documented come from the model, the tool, the mode and the question policy."}],"url":"https://paulantoinetual.fr/en/blog/negation-majuscules-questions-prompts-ce-que-la-preuve-soutient","markdownUrl":"https://paulantoinetual.fr/en/blog/negation-majuscules-questions-prompts-ce-que-la-preuve-soutient.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/negation-majuscules-questions-prompts-ce-que-la-preuve-soutient.a2ui.json","image":"https://paulantoinetual.fr/img/og/negation-majuscules-questions-prompts-ce-que-la-preuve-soutient-en.jpg"},{"slug":"harnais-implementation-ia-mesurer-ce-quil-voit","title":"We measure what our AI harness sees, not what it prevents","description":"16.6% of the 2,661 plans closed by our AI implementation agents had at least one verification failure recorded before closing. No interception rate is publishable, and 14 of our 48 guardrails leave no trace. What a harness lets us claim, what it does not, and the five errors we corrected in our own method.","category":"Junyr Method™","categoryKey":"methode","published":"2026-11-04","tags":["AI harness","agentic coding","AI agents","implementation agents","guardrails","observability","measurement","human in the loop","methodology","Junyr Method™"],"words":2242,"readingMinutes":11,"sources":4,"figures":3,"topics":{"agents":10,"methode":9,"outils":3},"faq":[{"question":"Does an AI implementation harness reduce defects?","answer":"We cannot say. Our harness records the failures it sees, but we have no counterfactual and no measurable causal link between a guardrail and a prevented defect, so no interception rate is publishable."},{"question":"What share of AI-built changes fail a check before closing?","answer":"16.6%: 443 of the 2,661 plans closed between August and October 2026 had at least one failed verification recorded. It is a lower bound, because recording covered only 49.6% of plans in August, and it is best read on September and October."},{"question":"Why is a harness's interception rate hard to measure?","answer":"Because four conditions are missing: a counterfactual, a measurable causal link, stable recording coverage and reliable links between bug reports and plans."},{"question":"What does a company need to record to measure its harness?","answer":"For each guardrail, the result, the cause, the run, the plan, the lane and the AI model, written to one table on a best-effort basis."},{"question":"Do AI models leave their authorised zone often?","answer":"We do not know. No table attributes a write outside the declared footprint to a specific lane, and the model that ran each lane is recorded nowhere."}],"url":"https://paulantoinetual.fr/en/blog/harnais-implementation-ia-mesurer-ce-quil-voit","markdownUrl":"https://paulantoinetual.fr/en/blog/harnais-implementation-ia-mesurer-ce-quil-voit.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/harnais-implementation-ia-mesurer-ce-quil-voit.a2ui.json","image":"https://paulantoinetual.fr/img/og/harnais-implementation-ia-mesurer-ce-quil-voit-en.jpg"},{"slug":"plateforme-agentique-centree-humain","title":"The augmented expert: agentic platforms must be built around people","description":"AI still falls short of a professional on four expert projects in five, and the human's judgement is what turns an agent's report into a deliverable. A benchmark of eleven agentic platforms in October 2026 shows the same blind spot: built for one person, for developers or for processes, they rarely give people a shared space to review, correct and sign the agents' work. Junyr's approach starts from that space.","category":"AI Agents","categoryKey":"agents","published":"2026-10-29","tags":["agentic AI","AI agents","augmented expert","human in the loop","human-agent interaction","agentic platforms","Microsoft Foundry","Agent 365","Claude","ChatGPT Work","ServiceNow","Agentforce","Remote Labor Index","Junyr"],"words":2460,"readingMinutes":12,"sources":21,"figures":3,"topics":{"agents":12,"marches":3,"outils":3},"faq":[{"question":"Does AI outperform human experts on professional work in 2026?","answer":"Not on complete professional deliverables. On the Remote Labor Index, which compares AI deliverables with those of experienced freelancers on 240 real projects, the best model in October 2026 matches or beats the professional on 21.25% of projects."},{"question":"Why should a person rework the reports that AI agents produce?","answer":"Because the person brings the context, the judgement and the accountability that the model lacks, and because a review that is not designed for tends to become a rubber stamp."},{"question":"What do today's agentic platforms get wrong?","answer":"They are built around a single user, around developers or around processes and the organisation chart, and they reduce the human to an approval step. A shared space where several people and agents work on the same objects, revise the deliverable and sign it is rarely documented."},{"question":"Is Microsoft Foundry suited to business users who supervise agents?","answer":"No. Microsoft describes Foundry as a platform that 'empowers developers', and its human-in-the-loop features are developer constructs, with no front end for a business user to supervise, correct or revise an agent's work."},{"question":"What is Junyr's approach to human-agent work?","answer":"Agents take the repetitive preparation, and people keep judgement, decisions and signatures, in one workspace where they see what the agents prepared and act on it."}],"url":"https://paulantoinetual.fr/en/blog/plateforme-agentique-centree-humain","markdownUrl":"https://paulantoinetual.fr/en/blog/plateforme-agentique-centree-humain.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/plateforme-agentique-centree-humain.a2ui.json","image":"https://paulantoinetual.fr/img/og/plateforme-agentique-centree-humain-en.jpg"},{"slug":"mots-cles-agents-ia-vocabulaire-consigne","title":"Keywords for AI agents: what the evidence supports, and the ten we teach","description":"Lists of trigger words for AI agents circulate with claims their sources do not make. We checked them against the papers and current vendor guidance, then set out the ten action and effort keywords Junyr teaches and where each word belongs: prompt, tool or harness.","category":"AI Agents","categoryKey":"agents","published":"2026-10-13","tags":["AI agents","prompt keywords","trigger words","prompt engineering","agent harness","stop conditions","sycophancy","MCP","Junyr Method","agent training","Claude","GPT-5.6"],"words":1622,"readingMinutes":8,"sources":10,"figures":3,"topics":{"agents":12,"methode":6,"prompts":10},"faq":[{"question":"Are there words that switch an AI agent into a rigorous mode?","answer":"Not in the sources we checked. No standard registry of such words exists, and the 2026 vendor guides ask for something different: a clear destination, the reason behind a rule, and a criterion the agent can check."},{"question":"Which keywords does Junyr teach?","answer":"Ten, in two families. Seven action words say what to do (AUDIT, EVAL, INVESTIGATE, TEST, REVIEW, ALIGN, HYGIENE) and three effort words say how hard to try (LIGHT, DEEP, GOLDEN STANDARD)."},{"question":"Should a stop condition be a word in the prompt?","answer":"A prompt can ask an agent to stop, but only the harness can guarantee it. Write the stop rule before launch and enforce it outside the prompt."},{"question":"Is the sentence 'Take a deep breath and work on this problem step-by-step' still worth using?","answer":"The 2023 DeepMind paper found it with an optimisation loop and measured it on one model of that time. The sources we checked say nothing about its effect on 2026 models, so test it on yours."}],"url":"https://paulantoinetual.fr/en/blog/mots-cles-agents-ia-vocabulaire-consigne","markdownUrl":"https://paulantoinetual.fr/en/blog/mots-cles-agents-ia-vocabulaire-consigne.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/mots-cles-agents-ia-vocabulaire-consigne.a2ui.json","image":"https://paulantoinetual.fr/img/og/mots-cles-agents-ia-vocabulaire-consigne-en.jpg"},{"slug":"agentic-coding-entreprises-construire-outil-de-gestion","title":"Agentic coding: why companies will build their own management tool","description":"When code was expensive, buying packaged software was the right call. As agentic coding makes code almost free, building the management core in-house becomes the better investment and turns the tool into an asset, while market data and distribution stay external.","category":"Junyr Method™","categoryKey":"methode","published":"2026-10-12","tags":["agentic coding","build or buy","custom software","CRM","ERP","full stack builders","UltraCoding","roadmap-driven development","information system","on-premise infrastructure","software as an asset","Junyr Method"],"words":969,"readingMinutes":5,"sources":0,"figures":3,"topics":{"agents":7,"methode":12,"outils":10},"faq":[{"question":"Why does agentic coding change the build-or-buy decision?","answer":"Because the cost of writing code collapses. The reasoning of build or buy stays the same: when buying is cheaper, you buy. What changes is that building a management tool tailored to the company now often costs less than buying one and adapting the organisation to it."},{"question":"Which tools will companies still buy?","answer":"Services that rest on data or reach the company cannot build alone: market data upstream, and distribution platforms downstream."},{"question":"What does an in-house management tool require from the company?","answer":"A small team of full stack builders working with AI coding, a structured method, and a person in charge of infrastructure. Junyr's estimate is one to three developers with support for the core tool."},{"question":"Should infrastructure stay on a cloud provider such as Azure?","answer":"Not automatically. Managing infrastructure can now largely be entrusted to an LLM agent, so the reasons that justified a cloud provider no longer necessarily hold."}],"url":"https://paulantoinetual.fr/en/blog/agentic-coding-entreprises-construire-outil-de-gestion","markdownUrl":"https://paulantoinetual.fr/en/blog/agentic-coding-entreprises-construire-outil-de-gestion.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/agentic-coding-entreprises-construire-outil-de-gestion.a2ui.json","image":"https://paulantoinetual.fr/img/og/agentic-coding-entreprises-construire-outil-de-gestion-en.jpg"},{"slug":"projets-agentiques-pas-de-roi-positif","title":"Why agentic projects don't (yet) have a positive ROI","description":"Agentic projects rarely pay back today, less because of the technology than because companies copy their human processes and organisation charts onto generalist models. What taking over failed projects taught us, and a three-step path: equip people, imagine the company, then give agents objectives rather than tasks.","category":"AI Agents","categoryKey":"agents","published":"2026-10-10","tags":["agentic AI","AI agents","AI ROI","AI transformation","process redesign","AI coworkers","generalist models","Claude Cowork","ChatGPT Work","forward deployed engineers","on-premise AI","agent objectives"],"words":2384,"readingMinutes":11,"sources":18,"figures":3,"topics":{"agents":12,"couts":6,"methode":3,"outils":3},"faq":[{"question":"Why don't agentic projects have a positive ROI yet?","answer":"Because most of them automate tasks inside processes designed for people, instead of rethinking how the company creates value for its clients. The agents work; the organisation they reproduce was never designed for them."},{"question":"Are specialised AI agents better than generalist ones?","answer":"Not on broad professional work so far. The models behind agents are generalists, and in a June 2026 Nature Medicine study three frontier general models outperformed two specialised clinical AI tools in all three evaluations."},{"question":"What should a company do before deploying agents?","answer":"Spend about six months equipping its teams with existing tools, Claude and ChatGPT connected to the company's software, while building a shared understanding of language models and designing the target organisation. Agents come afterwards, with objectives."},{"question":"Why give an agent a global objective rather than a local one?","answer":"Because an agent pursues the objective it is given and nothing else: a sales agent trades margin for volume, a finance agent cuts spending without commercial vision, and rules get worked around."},{"question":"Can an AI agent replace an experienced professional?","answer":"Not in the core of the job. After almost nine months of agentic AI in companies, we find that it assists with secondary tasks but does not replace know-how."},{"question":"What is a forward deployed engineer, and should a company hire them?","answer":"A forward deployed engineer (FDE) is sent into a customer's organisation to configure AI tools, integrate software and solve business problems on site. They only create value if what they deploy has been rethought first."},{"question":"Why should a company own its agentic platform?","answer":"Because a platform built on a vendor's product follows the vendor's calendar: models retired, a harness changed underneath, prices decided elsewhere. A platform hosted on-premises, with the models the company chose, is updated when the company decides."}],"url":"https://paulantoinetual.fr/en/blog/projets-agentiques-pas-de-roi-positif","markdownUrl":"https://paulantoinetual.fr/en/blog/projets-agentiques-pas-de-roi-positif.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/projets-agentiques-pas-de-roi-positif.a2ui.json","image":"https://paulantoinetual.fr/img/og/projets-agentiques-pas-de-roi-positif-en.jpg"},{"slug":"la-guerre-des-prix-a-commence","title":"The price war has begun","description":"Opus 5.5 cuts its price list by 20%, GPT-6 Sol halves its own, and for most business work the intelligence gap between frontier models is no longer visible. What the competition now runs on, what an agent task really costs, and where the frontier model still earns its price.","category":"Markets & AI","categoryKey":"marches","published":"2026-09-27","tags":["AI pricing","price war","Claude Opus 5.5","GPT-6","prompt caching","Jevons paradox","agent harness","system prompt","cost per task","prompt injection","frontier models","Junyr Method"],"words":3220,"readingMinutes":15,"sources":20,"figures":3,"topics":{"agents":4,"couts":10,"marches":6,"methode":3,"prompts":6},"faq":[{"question":"Did Anthropic cut the price of Claude Opus 5.5 by 40%?","answer":"No. The price list fell by 20%, to $4 and $20 per million input and output tokens against $5 and $25 for Opus 5. The 40% is Anthropic's estimate of what a typical workload costs to run at default settings."},{"question":"Why can users no longer see an intelligence difference between frontier models?","answer":"Because at this level, configuration and style move the result more than the generation of the model, and the vendors now say so themselves."},{"question":"How much of a model's perceived behaviour comes from its system prompt?","answer":"In the chat apps, a measurable part. Anthropic publishes the claude.ai system prompt for each model, and the text changes from one model to the next."},{"question":"Why does the AI bill rise when the price of the token falls?","answer":"It is Jevons's paradox: a cheaper resource gets used for more things, and agent workflows consume several orders of magnitude more tokens than a question and its answer."},{"question":"How much does prompt caching change the cost of an agent task?","answer":"More than the price list does. On an illustrative agent task, the share of input served from the cache moves the cost by a factor of more than five."},{"question":"When is a frontier model still worth its price?","answer":"In three cases: code on large codebases, agents that read untrusted content, and autonomous decisions that cannot be undone."},{"question":"What should a company measure instead of asking which model is best?","answer":"The cost of one case file processed end to end, reliably, and without depending on the moods of a single supplier."}],"url":"https://paulantoinetual.fr/en/blog/la-guerre-des-prix-a-commence","markdownUrl":"https://paulantoinetual.fr/en/blog/la-guerre-des-prix-a-commence.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/la-guerre-des-prix-a-commence.a2ui.json","image":"https://paulantoinetual.fr/img/og/la-guerre-des-prix-a-commence-en.jpg"},{"slug":"orchestration-frugale-agents-ia","title":"Lean orchestration: five rules from the bill of a multi-agent run","description":"On one multi-agent development run, 19% of the cost went to subagent caches that expired while tests ran, and 8% to starting new agents. Five rules on waiting, spawning, reading, reporting and orchestrating, with the prices and Claude Code settings behind each one.","category":"AI Agents","categoryKey":"agents","published":"2026-09-27","updated":"2026-10-11","tags":["lean orchestration","multi-agent systems","subagents","prompt caching","cache TTL","Claude Code","Claude Opus 5.5","Claude Fable 5.1","context engineering","context compaction","AI cost","agent harness"],"words":2697,"readingMinutes":13,"sources":18,"figures":3,"topics":{"agents":12,"couts":4,"prompts":7},"faq":[{"question":"Why does a subagent's cache expire during a long test run?","answer":"Because Claude Code gives subagents a five-minute cache and the clock keeps running while a tool call blocks, so a wait of more than about four minutes forces the next request to write the whole context again."},{"question":"How much does a cache expiry cost compared with waiting in slices?","answer":"On a context of 200,000 tokens, a rewrite costs $1.00 on Opus 5.5 and $2.50 on Fable 5.1, while a four-minute waiting slice that reads the cache costs about $0.04 and $0.05."},{"question":"Should subagents simply be given a one-hour cache?","answer":"Sometimes. Claude Code has allowed it since version 2.1.242, but a one-hour cache write costs twice the input price against 1.25 times for five minutes, so it only pays for agents likely to wait more than five minutes."},{"question":"Why reuse a warm agent instead of spawning a new one?","answer":"Because each new agent writes its whole starting context, about 90,000 tokens on this run, while a warm agent reads the same prefix from the cache for a twenty-fifth of the price on Opus 5.5."},{"question":"Why should the implementer not verify its own work?","answer":"Because models judge their own output too kindly, and because the implementer's context is full of failed attempts that make verification slower and more expensive."},{"question":"How hard should tool output be filtered before it enters an agent's context?","answer":"Hard, with the full output kept on disk: most of a test log or a file is noise, but aggressive filtering also drops a share of the useful lines."},{"question":"Why keep the orchestrator's context small when the model has a million-token window?","answer":"Because the orchestrator re-reads its whole context on every response, so its size multiplies the cost of each of its thousands of turns."},{"question":"Which model should run the orchestrator?","answer":"Not necessarily the most expensive one. Most of an orchestrator's cost is re-reading its context, and Claude Sonnet 5.5 or GPT-6.1 Sol at xhigh halve that price against Opus 5.5, provided the project runs a strict development flow."}],"url":"https://paulantoinetual.fr/en/blog/orchestration-frugale-agents-ia","markdownUrl":"https://paulantoinetual.fr/en/blog/orchestration-frugale-agents-ia.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/orchestration-frugale-agents-ia.a2ui.json","image":"https://paulantoinetual.fr/img/og/orchestration-frugale-agents-ia-en.jpg"},{"slug":"ingenierie-harnais-agent-ia","title":"Harness engineering: the software layer that turns a model into an agent","description":"What an agent harness is, what a source-code audit of eleven production systems actually found inside one, and which of the figures circulating about agent failure can be sourced.","category":"AI Agents","categoryKey":"agents","published":"2026-09-20","tags":["agent harness","harness engineering","AI agents","context engineering","prompt injection","Model Context Protocol","MCP security","software supply chain","sandboxing","context rot","agentic architecture","Junyr Method"],"words":4497,"readingMinutes":21,"sources":38,"figures":3,"topics":{"agents":12,"souverainete":3,"methode":3,"prompts":6,"outils":6},"faq":[{"question":"What is an agent harness?","answer":"The harness is everything in an agentic system except the model: the runtime that couples a language model to the world through a loop, a tool interface, context management and control mechanisms."},{"question":"How is harness engineering different from prompt engineering and context engineering?","answer":"The three tiers differ by what you are allowed to change: the wording of one instruction, the contents of the attention window, or the environment the model acts in."},{"question":"Does a better harness beat a better model?","answer":"Sometimes, and it is measurable, but the honest version is narrower than the slogan: published gains come from named techniques on named benchmarks, not from a general law."},{"question":"Why should the part of the agent that reasons not be the part that acts?","answer":"Because an instruction hidden in retrieved content reaches the reasoning step, so any privilege the reasoning step holds is a privilege the attacker inherits."},{"question":"What actually goes wrong in production agent systems?","answer":"The best-evidenced answer is a taxonomy built from annotated execution traces, which finds most failures upstream of the model's raw capability, in specification, coordination and verification."},{"question":"Is it true that 88% of agent projects fail and that 65% of the failures are infrastructure?","answer":"No source supports either figure. Both circulate through blog posts that cite nothing, and the sourced figures on the same question are different numbers measuring different things."},{"question":"Should MCP servers count as part of the harness, or as a separate term?","answer":"Every published definition folds them into the harness. This article argues they deserve a separate line on a different axis, because a third-party server is the only component that can change under you without a version number."},{"question":"What should a mid-sized company do with this?","answer":"Treat the harness as the part you own and version, because the model is the part you rent and cannot control."}],"url":"https://paulantoinetual.fr/en/blog/ingenierie-harnais-agent-ia","markdownUrl":"https://paulantoinetual.fr/en/blog/ingenierie-harnais-agent-ia.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/ingenierie-harnais-agent-ia.a2ui.json","image":"https://paulantoinetual.fr/img/og/ingenierie-harnais-agent-ia-en.jpg"},{"slug":"pause-ia-modele-economique-labos-frontiere","title":"The AI pause is also a business model: what frontier labs gain by slowing down, and what the other 90% leaves to your company","description":"Why frontier labs can afford to slow down, what the 10% extinction figure really is, and what the remaining 90% leaves for a mid-sized company to build.","category":"Markets & AI","categoryKey":"marches","published":"2026-09-17","tags":["AI pause","frontier labs","OpenAI","Anthropic","training costs","inference margins","model lifecycle","existential risk","AI governance","work design","productivity","mid-sized companies"],"words":2639,"readingMinutes":13,"sources":24,"figures":3,"topics":{"couts":6,"marches":9,"methode":6,"productivite":3},"faq":[{"question":"Did OpenAI and Anthropic stop training models in August and September 2026?","answer":"No. OpenAI paused reinforcement learning on its deployment models for two weeks and kept its largest frontier run on hold, while Anthropic's chief executive wrote that pacing 'does not mean halting model training'."},{"question":"Why would a slower release cadence improve a frontier lab's finances?","answer":"Because the cash leaves through training the next model and comes back through serving the current one, so each extra month at the top of the price list earns at inference margins above 80% without paying for a new run."},{"question":"Who said there is a ten percent chance that AI kills everyone within a decade?","answer":"Evan Hubinger, Anthropic's alignment science lead, in a post on X supporting the resignation of researcher Jacob Coxon on 8 September 2026. It is a personal estimate, not a consensus figure."},{"question":"Is a ten percent risk acceptable because the other ninety percent is favourable?","answer":"No. A one-in-ten systemic failure rate would be intolerable in any industry, and the article says so. The ninety percent is not a reassurance; it is the scenario a company can plan for."},{"question":"Does AI actually relieve employees, or does it add load?","answer":"Both outcomes are measured in 2026, and the surveys agree that the difference comes from how the organisation designs the work rather than from the tool."},{"question":"What should a mid-sized company do with a slower frontier?","answer":"Treat it as the stable base it needed anyway: pin a model for a year, keep evaluation sets and prompts valid, and spend the saved requalification effort on work design."},{"question":"Is this article investment advice?","answer":"No. It reads published financial figures to explain an industry decision and offers no view on any security."}],"url":"https://paulantoinetual.fr/en/blog/pause-ia-modele-economique-labos-frontiere","markdownUrl":"https://paulantoinetual.fr/en/blog/pause-ia-modele-economique-labos-frontiere.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/pause-ia-modele-economique-labos-frontiere.a2ui.json","image":"https://paulantoinetual.fr/img/og/pause-ia-modele-economique-labos-frontiere-en.jpg"},{"slug":"devis-interactif-nouveau-devis","title":"The interactive quote is the new quote: what behavioural research says about simulators, co-creation and price transparency","description":"Why an AI-native company turns its quote into a simulator, and what twenty years of behavioural research say about how a buyer decides in front of one.","category":"2027 AI-native Company Insights","categoryKey":"ianative","published":"2026-09-14","tags":["interactive quote","CPQ","pricing","behavioural economics","IKEA effect","operational transparency","cost transparency","price partitioning","mental accounting","smart defaults","AI-native company","B2B sales"],"words":3272,"readingMinutes":15,"sources":26,"figures":3,"topics":{"couts":9,"outils":7},"faq":[{"question":"Does an interactive quote mean the customer sets their own price?","answer":"No: the buyer adjusts scope, volumes and options inside rules the seller keeps, and the price follows from that configuration rather than from a negotiation over the number."},{"question":"Is the IKEA effect strong enough to matter in a business-to-business purchase?","answer":"The original measurement is consumer-side: builders paid about 63% more for a storage box they had assembled themselves. The transfer to a business buyer is plausible but not measured, and it carries the same boundary condition, the configuration has to complete successfully."},{"question":"Should a quote engine deliberately slow down its answer?","answer":"Only if the pause shows real work: the labour illusion studies found people valued a slower site that displayed its operations above an instant one with identical results, while an unexplained wait behaved like a defect."},{"question":"Does revealing our costs mean giving away our margin?","answer":"The preregistered field experiment found a 21.1% rise in purchase probability when a cafeteria showed the cost of a bowl of soup next to its price, at a 17% gross margin, and the effect held at a 55% margin in a later study; whether it survives extreme markups or luxury positioning was never tested."},{"question":"Should a configurator start empty or fully loaded?","answer":"Neither extreme: a subtractive start (fully loaded) produces a higher total because removing an option feels like a loss, but a professional buyer can read that as manipulation, so the best-documented compromise is a profiled smart default with free adjustment in both directions."}],"url":"https://paulantoinetual.fr/en/blog/devis-interactif-nouveau-devis","markdownUrl":"https://paulantoinetual.fr/en/blog/devis-interactif-nouveau-devis.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/devis-interactif-nouveau-devis.a2ui.json","image":"https://paulantoinetual.fr/img/og/devis-interactif-nouveau-devis-en.jpg"},{"slug":"norme-securite-gouv-fr-post-quantique","title":"France's unwritten .gouv.fr security standard: no rate limiting, no intrusion detection, no MFA, and a compulsory post-quantum deadline","description":"What 2026's French state data breaches reveal as the real security floor, and why post-quantum migration just became a dated, compulsory deadline.","category":"Cybersecurity","categoryKey":"cyber","published":"2026-09-08","tags":["ANSSI","cybersecurity","data breach","post-quantum","PQC","MFA","intrusion detection","patch management","AI","public sector","CrowdStrike","France"],"words":1894,"readingMinutes":9,"sources":16,"figures":3,"topics":{"souverainete":12,"methode":3},"faq":[{"question":"Is there an official .gouv.fr security standard that actually says no rate limiting, no intrusion detection and no MFA?","answer":"No: no such document exists. The phrase describes what 2026's run of French public-sector breaches revealed in practice as the operating floor, not a published requirement."},{"question":"Does the 2027 ANSSI deadline mean every French SME must adopt post-quantum cryptography by then?","answer":"No: ANSSI's own FAQ targets 2027 for bringing PQC obligations into product qualification, not a blanket compliance deadline for every organisation."},{"question":"If the ANTS breach was caused by a broken access control flaw, would rate limiting have prevented it?","answer":"Not on its own: the entry point was an insecure direct object reference, a coding flaw that rate limiting does not fix by itself."},{"question":"Does using AI for security automatically lower breach costs?","answer":"No: the World Economic Forum's 2026 findings apply to organisations that already use AI extensively and deliberately in security operations, not to AI adoption in general."},{"question":"Is a daily reboot of production systems itself a security measure?","answer":"No: a restart with no patch, secret rotation or configuration change behind it is uptime housekeeping, not hygiene."}],"url":"https://paulantoinetual.fr/en/blog/norme-securite-gouv-fr-post-quantique","markdownUrl":"https://paulantoinetual.fr/en/blog/norme-securite-gouv-fr-post-quantique.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/norme-securite-gouv-fr-post-quantique.a2ui.json","image":"https://paulantoinetual.fr/img/og/norme-securite-gouv-fr-post-quantique-en.jpg"},{"slug":"processus-agentiques-entreprise","title":"Agentic processes in business: where code, skills and judgement belong","description":"A technical method for making an agentic process replayable: measure skill discovery, reserve judgement for useful steps and enforce controls in code.","category":"AI Agents","categoryKey":"agents","published":"2026-08-30","updated":"2026-09-06","tags":["agentic processes","AI agents","Agent Skills","agentic workflow","software engineering","orchestration","determinism","MCP","human-in-the-loop","Junyr Method"],"words":1948,"readingMinutes":9,"sources":14,"figures":0,"topics":{"agents":12,"methode":4,"outils":3},"faq":[{"question":"What is the difference between a skill and an agentic workflow?","answer":"A skill makes know-how reusable by an agent, while a workflow makes the process sequence explicit, testable and resumable."},{"question":"Should a business agentic process be fully autonomous?","answer":"The appropriate level of autonomy depends on consequences, reversibility and the quality of controls rather than on a uniform goal of full autonomy."},{"question":"How should the reliability of an agentic chain be estimated?","answer":"Teams need conditional probabilities measured on representative cases, because multiplying stage rates is valid only under an independence assumption."},{"question":"What does software engineering add to an agentic process?","answer":"Software engineering turns a sequence of model calls into a system that can be observed, repaired and governed."},{"question":"Where should the boundary between code and agent sit?","answer":"The boundary should be decided step by step according to verifiability, the nature of the work and the consequence of the action."},{"question":"Are Agent Skills portable?","answer":"Agent Skills is published as an open standard and several vendors support it, but practical portability still depends on each client's tools, permissions and conventions."},{"question":"Where should an SME with no agentic process begin?","answer":"Begin with a frequent or periodic process that is documented, measurable and reversible, then map its steps before selecting models."}],"url":"https://paulantoinetual.fr/en/blog/processus-agentiques-entreprise","markdownUrl":"https://paulantoinetual.fr/en/blog/processus-agentiques-entreprise.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/processus-agentiques-entreprise.a2ui.json","image":"https://paulantoinetual.fr/img/og/processus-agentiques-entreprise-en.jpg"},{"slug":"tiret-cadratin-signature-ecriture-ia","title":"Em dashes and AI writing: a style clue, never proof","description":"What dash frequency really reveals: editorial habits, French and English conventions, model measurements and the limits of individual detection.","category":"Personal Effectiveness","categoryKey":"efficacite","published":"2026-08-28","updated":"2026-09-06","tags":["generative writing","em dash","typography","stylometry","RLHF","alignment","tokenisation","BPE","AI text detection","French typography","contrastive rhetoric"],"words":1535,"readingMinutes":7,"sources":8,"figures":0,"topics":{"couts":3,"prompts":9,"productivite":3},"faq":[{"question":"Why do some models use so many em dashes?","answer":"The available measurements primarily show variation between models and prompts, while the explanation involving Markdown and post-training remains a hypothesis supported by one unreplicated preprint."},{"question":"Does an em dash prove that AI wrote a text?","answer":"No, because dash frequency overlaps widely between human authors and models, and the available studies explicitly reject the character as a detector for individual texts."},{"question":"How can you reduce dashes in AI-assisted text?","answer":"Specify the required typographic convention, then review or count the characters, because compliance varies with the model, prompt and output length."},{"question":"What is the French typographic convention for dashes?","answer":"French typography distinguishes em and en dashes, although their use in parenthetical clauses varies between style guides and publishers."},{"question":"Where does the balanced sentence in generated prose come from?","answer":"Research on diverse preferences, output diversity and sycophancy makes this form plausible, but none of the cited papers directly measures its frequency or cause."},{"question":"Should writers ban em dashes?","answer":"The choice depends on the author's voice and house style, because a well-used dash remains legitimate punctuation and removing every instance does not improve every sentence."}],"url":"https://paulantoinetual.fr/en/blog/tiret-cadratin-signature-ecriture-ia","markdownUrl":"https://paulantoinetual.fr/en/blog/tiret-cadratin-signature-ecriture-ia.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/tiret-cadratin-signature-ecriture-ia.a2ui.json","image":"https://paulantoinetual.fr/img/og/tiret-cadratin-signature-ecriture-ia-en.jpg"},{"slug":"cowork-agents-ia-durables","title":"Cowork and durable agents: regaining control of dependencies","description":"Cowork can support agent discovery and some production work, but durability depends on testing, contracts, controlled versions and a fallback plan.","category":"AI Agents","categoryKey":"agents","published":"2026-08-27","updated":"2026-09-06","tags":["AI agents","agentic design","Claude Cowork","software harness","safety classifier","model deprecation","availability","AI governance","agentic engineering","Junyr Method"],"words":1369,"readingMinutes":6,"sources":10,"figures":0,"topics":{"agents":12,"methode":6,"outils":3},"faq":[{"question":"Why is Claude Cowork insufficient, on its own, to guarantee a durable business agent?","answer":"Cowork can support durable use when its dependencies and limits suit the process, but the product does not transfer control of every layer that determines its behaviour to the business."},{"question":"What is an AI agent harness?","answer":"The harness is the software layer that turns a model into a usable agent, and changes to it can alter a process even when the model, data and business instruction stay the same."},{"question":"Does a Claude model ID guarantee perfectly stable behaviour?","answer":"According to Anthropic, a Claude model ID pins the model's weights and configuration, but that guarantee covers neither every serving layer nor the model's indefinite availability."},{"question":"Does the MCP protocol automatically make a connector stable?","answer":"MCP makes the interface inspectable and versioned, but connector stability still depends on the negotiated version, the server, its permissions and the third-party system it calls."},{"question":"Does Anthropic guarantee availability for Cowork or the Claude API?","answer":"Anthropic's general commercial terms do not promise uninterrupted service, while any negotiated commitments and applicable service levels must be read in each customer's own contract."},{"question":"Is this continuity risk specific to Anthropic?","answer":"Dependence on a managed service affects every supplier because models, APIs, connectors, prices and retirement policies evolve on schedules that customers do not control completely."},{"question":"What should a business build around Cowork for durable use?","answer":"The right target is a reversible, observable process whose essential controls remain the business's responsibility, whether execution relies on Cowork, an API or an internally hosted model."}],"url":"https://paulantoinetual.fr/en/blog/cowork-agents-ia-durables","markdownUrl":"https://paulantoinetual.fr/en/blog/cowork-agents-ia-durables.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/cowork-agents-ia-durables.a2ui.json","image":"https://paulantoinetual.fr/img/og/cowork-agents-ia-durables-en.jpg"},{"slug":"le-chat-nest-pas-linterface-de-travail","title":"After chat: designing the AI work interface","description":"Chat remains useful for exploration and exceptions, but durable processes need business context, controls and an interface fitted to the action.","category":"AI Agents","categoryKey":"agents","published":"2026-08-27","updated":"2026-09-06","tags":["AI agents","user interfaces","Claude Cowork","generative UI","AG-UI","Model Context Protocol","human in the loop","Service-as-Software","Junyr Method"],"words":781,"readingMinutes":4,"sources":7,"figures":2,"topics":{"agents":8,"methode":3,"prompts":4,"outils":10},"faq":[{"question":"Is chat useless for working with AI?","answer":"Chat remains a strong surface for exploring a subject, expressing an unusual request or learning, while repeated tasks benefit from a structured business interface."},{"question":"Why can a business interface be more reliable than chat?","answer":"A business interface supplies some of the context and constraints that a user would otherwise need to restate in every exchange."},{"question":"What is a generative interface?","answer":"A generative interface adapts its presentation and components to the intent, without assuming that every interaction should end as a text response."},{"question":"Does AG-UI replace MCP?","answer":"AG-UI and MCP cover different connections and can work together in the same agentic application."},{"question":"How should a team decide where to keep chat?","answer":"The choice should follow task variability, risk and the context that the interface can supply automatically."}],"url":"https://paulantoinetual.fr/en/blog/le-chat-nest-pas-linterface-de-travail","markdownUrl":"https://paulantoinetual.fr/en/blog/le-chat-nest-pas-linterface-de-travail.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/le-chat-nest-pas-linterface-de-travail.a2ui.json","image":"https://paulantoinetual.fr/img/og/le-chat-nest-pas-linterface-de-travail-en.jpg"},{"slug":"decider-avec-ses-donnees-ia-agentique","title":"Deciding with your data: what agentic AI really changes","description":"A practical method for framing, analysing and checking a decision with company data, while separating compute cost, integration cost and reliability.","category":"AI Agents","categoryKey":"agents","published":"2026-08-18","updated":"2026-09-06","tags":["agentic AI","decision support","data analysis","projection scenarios","cognitive bias","intuition","MCP","BI","data analyst","proprietary data","text-to-SQL","governance"],"words":1760,"readingMinutes":9,"sources":27,"figures":3,"topics":{"agents":11,"methode":4},"faq":[{"question":"What can an agentic AI connected to my company's data actually analyse?","answer":"An agent can combine the data it is authorised to reach and prepare a decision dossier, provided that business definitions, permissions and checks are explicit."},{"question":"How much does an AI data analysis cost in 2026?","answer":"The compute for one analysis may cost only a few euros, while the full cost depends mainly on data preparation, integration, security and review time."},{"question":"Does AI analysis replace the business leader's intuition?","answer":"Intuition supplies a useful hypothesis, while agentic analysis can test it against observable facts before the leader takes responsibility for the choice."},{"question":"Can the figures produced by an AI agent be trusted?","answer":"Figures become decision-grade when their provenance, calculation and reconciliation against a system of record can be reproduced."},{"question":"Does an AI agent replace a data analyst or a BI project?","answer":"An agent automates part of extraction and explanation, while analysts and data teams remain responsible for meaning, quality and the control framework."},{"question":"My competitors use the same AIs: where is the advantage?","answer":"The advantage depends less on the chosen model than on private data quality, business rules and the learning speed of your decision loop."},{"question":"Can AI take the decision for me?","answer":"AI can execute bounded operational decisions, but the organisation must assign clear accountability for choices that materially commit the business."}],"url":"https://paulantoinetual.fr/en/blog/decider-avec-ses-donnees-ia-agentique","markdownUrl":"https://paulantoinetual.fr/en/blog/decider-avec-ses-donnees-ia-agentique.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/decider-avec-ses-donnees-ia-agentique.a2ui.json","image":"https://paulantoinetual.fr/img/og/decider-avec-ses-donnees-ia-agentique-en.jpg"},{"slug":"la-fin-de-la-course-au-modele","title":"Choosing an AI model without betting on the next champion","description":"Rankings move, benchmarks age and prices fall: a practical method for comparing AI models on quality, cost, risk and reversibility.","category":"Markets & AI","categoryKey":"marches","published":"2026-08-17","updated":"2026-09-06","tags":["AI","commoditisation","frontier models","open weights","benchmarks","LMArena","API costs","reversibility","evaluation"],"words":1290,"readingMinutes":6,"sources":9,"figures":0,"topics":{"couts":4,"marches":8},"faq":[{"question":"Is the model race really over?","answer":"Technical competition continues, but its current ranking is no longer enough to guide a durable investment."},{"question":"Can a public leaderboard identify the best model?","answer":"A public leaderboard provides a useful signal, but it measures only part of the service a business needs."},{"question":"Have open-weight models caught up with closed models?","answer":"They have moved closer on several public evaluations without making the two delivery models interchangeable."},{"question":"Why is the price per million tokens insufficient?","answer":"The useful measure is the cost of an accepted production task, rather than the price of a text unit in isolation."},{"question":"Should I wait for the next model before launching a project?","answer":"Waiting is rarely justified when the need is clear and the architecture allows the supplier to be replaced."},{"question":"Should a business fine-tune an open model on its data?","answer":"Fine-tuning becomes relevant when a stable, repetitive and measurable domain resists simpler approaches."},{"question":"Is this article investment advice?","answer":"This article offers a technical purchasing and evaluation method without recommending any financial security."}],"url":"https://paulantoinetual.fr/en/blog/la-fin-de-la-course-au-modele","markdownUrl":"https://paulantoinetual.fr/en/blog/la-fin-de-la-course-au-modele.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/la-fin-de-la-course-au-modele.a2ui.json","image":"https://paulantoinetual.fr/img/og/la-fin-de-la-course-au-modele-en.jpg"},{"slug":"charge-mentale-ia-sans-perdre-son-jugement","title":"Cutting your mental load with AI: the personal-effectiveness method","description":"Decide in the morning, let agents execute in the afternoon, dictate to emit, read to approve: the personal-effectiveness techniques, with the measurements behind them.","category":"Personal Effectiveness","categoryKey":"efficacite","published":"2026-08-07","updated":"2026-09-06","tags":["mental load","personal effectiveness","asynchronous work","AI agents","delegation","team management","automation bias","attention residue","voice dictation","Junyr Method"],"words":2284,"readingMinutes":11,"sources":14,"figures":3,"topics":{"agents":4,"methode":9,"productivite":11},"faq":[{"question":"Does AI actually save time?","answer":"Time savings depend on the task, user and study design, while a 2026 experiment mainly found lower perceived effort on simple cognitive tasks."},{"question":"Is lowering your mental load a real gain if the clock does not move?","answer":"Lower perceived effort can be useful without measured acceleration, provided it is tracked as a separate outcome and is not automatically credited with higher productivity or better decisions."},{"question":"What is a \"decision session\"?","answer":"A decision session is a protected window in which you batch consequential judgements and turn each one into a written mandate before agents begin execution."},{"question":"Why concentrate decisions in the morning?","answer":"There is no established general cognitive advantage to the morning; the practical benefit comes from choosing a genuinely protectable window that reduces switching among decision, execution and review."},{"question":"How is supervising agents a matter of team management?","answer":"Supervising agents uses familiar management practices, with controls suited to systems that can produce quickly and confidently without understanding the consequences."},{"question":"Should you dictate to AI or type to it?","answer":"Dictation can speed up the capture of an intention, although the available measurements mostly cover transcription of known text and do not prove an advantage for composing new thought."},{"question":"And for output: text, audio, diagram or video?","answer":"For a decision that must be checked, text and diagrams support backtracking and comparison, while audio and video are better suited to sequential presentation."},{"question":"Where does a mid-sized company start with this method?","answer":"A smaller company can test the method on one workflow for a week, using one decision window, a written mandate and a batched review, then retain what improves its own measures."}],"url":"https://paulantoinetual.fr/en/blog/charge-mentale-ia-sans-perdre-son-jugement","markdownUrl":"https://paulantoinetual.fr/en/blog/charge-mentale-ia-sans-perdre-son-jugement.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/charge-mentale-ia-sans-perdre-son-jugement.a2ui.json","image":"https://paulantoinetual.fr/img/og/charge-mentale-ia-sans-perdre-son-jugement-en.jpg"},{"slug":"saaspocalypse-10000-logiciels-par-jour","title":"Saaspocalypse: abundant prototypes are changing the value of SaaS","description":"GitHub repositories, generated projects and maintained products cannot be pooled: what their growth really changes for SaaS, procurement and governance.","category":"Markets & AI","categoryKey":"marches","published":"2026-08-02","updated":"2026-09-06","tags":["SaaS","Saaspocalypse","vibe coding","Claude Code","Lovable","GitHub","per-seat pricing","software market","generative AI","Airtable","Bending Spoons","Junyr Method"],"words":1390,"readingMinutes":6,"sources":10,"figures":0,"topics":{"couts":3,"marches":12,"methode":4,"outils":7},"faq":[{"question":"What is the 'Saaspocalypse'?","answer":"The term describes the repricing of SaaS vendors as agents reduce the cost of some tasks and weaken per-seat billing, without implying that the whole sector will disappear."},{"question":"Can we say that 10,000 new pieces of software are created every day?","answer":"Daily volumes above 10,000 can be documented for some creation artefacts, but none of the cited sources measures 10,000 distinct software products that are published, operated and maintained each day."},{"question":"What is seat compression?","answer":"Seat compression is the possible reduction in human licences when an agent performs part of the work, although its scale depends on the process, the contract and the degree of control required."},{"question":"Does the Veracode benchmark give the probability that an application is vulnerable?","answer":"No: its 45% rate describes security failures observed in a defined benchmark of tasks, languages and vulnerability classes, rather than a calibrated probability for an arbitrary production application."},{"question":"What does the Airtable acquisition agreement announced on 4 August 2026 show?","answer":"The agreement provides a negotiated price for Airtable and confirms that Hyperagent had been carved out before signing, but it proves neither that AI alone explains the price nor that every SaaS business will follow the same path."},{"question":"Is this article investment advice?","answer":"No: this article analyses business models and public sources for educational purposes, without recommending the purchase or sale of any security."}],"url":"https://paulantoinetual.fr/en/blog/saaspocalypse-10000-logiciels-par-jour","markdownUrl":"https://paulantoinetual.fr/en/blog/saaspocalypse-10000-logiciels-par-jour.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/saaspocalypse-10000-logiciels-par-jour.a2ui.json","image":"https://paulantoinetual.fr/img/og/saaspocalypse-10000-logiciels-par-jour-en.jpg"},{"slug":"developpement-a-decision-gardee","title":"Decision-gated development: approve the plan before writing code","description":"A method for working with AI agents that separates exploration, human decisions and implementation, illustrated by seventeen days from Junyr's internal registry.","category":"Junyr Method™","categoryKey":"methode","published":"2026-07-29","updated":"2026-09-14","tags":["roadmap-driven development","AI agents","agentic engineering","human-in-the-loop","AI governance","UltraCoding","decision-complete plans","Junyr Method","spec-driven development","outcome-driven prioritization","end-user outcome"],"words":1625,"readingMinutes":8,"sources":4,"figures":3,"topics":{"agents":7,"methode":9,"prompts":3,"outils":9},"faq":[{"question":"What is decision-gated development?","answer":"Decision-gated development formalises a decision in a reviewable plan, then limits implementation to work that a human has explicitly approved."},{"question":"Why does planning parallelise while implementation does not?","answer":"Several explorations can proceed in parallel while they avoid shared state, whereas code changes must be coordinated around the repository's actual dependencies."},{"question":"What is a \"decision-complete\" plan?","answer":"A decision-complete plan contains the choices needed to begin without reopening the underlying analysis, while leaving mechanical details to implementation."},{"question":"What is the method's actual yield?","answer":"Junyr's seventeen-day internal registry primarily shows that its gate stopped some work before code, without establishing a yield that can be generalised to other teams."},{"question":"What is outcome-driven prioritization?","answer":"Outcome-driven prioritization ranks the approved plans of the next wave by the effect expected for end users, corrected by the effect observed after the previous wave, without bypassing the human gate."},{"question":"How does this method relate to the Junyr Method™?","answer":"Decision-gated development applies to engineering work the same principle of explicit responsibility that the Junyr Method™ applies to organisational transformation."}],"url":"https://paulantoinetual.fr/en/blog/developpement-a-decision-gardee","markdownUrl":"https://paulantoinetual.fr/en/blog/developpement-a-decision-gardee.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/developpement-a-decision-gardee.a2ui.json","image":"https://paulantoinetual.fr/img/og/developpement-a-decision-gardee-en.jpg"},{"slug":"bulle-speculative-ia-pivot-mistral","title":"AI investment and sovereignty: the Microsoft-Mistral partnership","description":"A dated reading of tensions in the AI cycle and the sovereignty choices opened by the Microsoft-Mistral partnership announced on 21 July 2026.","category":"Markets & AI","categoryKey":"marches","published":"2026-07-23","updated":"2026-09-06","tags":["AI","speculative bubble","Nvidia","Mistral AI","Microsoft","capex","Minsky","digital sovereignty","financial literacy"],"words":1408,"readingMinutes":6,"sources":10,"figures":0,"topics":{"souverainete":7,"marches":12,"outils":3},"faq":[{"question":"Is AI going through its Minsky moment?","answer":"The Minsky-Kindleberger framework helps interpret the tensions observed in 2026, but it cannot date a market peak or predict a crash."},{"question":"What does Bank of America's July 2026 survey say about AI bubble risk?","answer":"Figures reported in the press suggest that fund managers had become more concerned, without turning their opinion into a market forecast."},{"question":"What does the 21 July 2026 Microsoft-Mistral announcement contain?","answer":"The official release describes a technology and commercial expansion around Mistral models, Microsoft environments and new Nvidia capacity in Europe."},{"question":"Is Mistral AI losing its French anchoring?","answer":"Public information shows a French company financed by international capital, but does not establish its effective control with certainty."},{"question":"Is the gap between AI spending and returns widening?","answer":"The evidence shows massive investment and uneven returns, without establishing one measured deficit for the entire ecosystem."},{"question":"Is this article investment advice?","answer":"This article offers an educational framework dated to 2026 and is neither investment advice nor a recommendation concerning any security."}],"url":"https://paulantoinetual.fr/en/blog/bulle-speculative-ia-pivot-mistral","markdownUrl":"https://paulantoinetual.fr/en/blog/bulle-speculative-ia-pivot-mistral.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/bulle-speculative-ia-pivot-mistral.a2ui.json","image":"https://paulantoinetual.fr/img/og/bulle-speculative-ia-pivot-mistral-en.jpg"},{"slug":"cac-40-vs-nasdaq-top-10","title":"CAC 40 and Nasdaq Top 10: what size does not tell you about AI maturity","description":"A critical reading of a 2026 study of publicly documented AI maturity across two large-cap panels drawn from the CAC 40 and Nasdaq-100.","category":"Markets & AI","categoryKey":"marches","published":"2026-07-20","updated":"2026-09-06","tags":["CAC 40","Nasdaq","market capitalisation","AI maturity","AI governance","AI Act","index concentration","NVIDIA","LVMH","financial literacy"],"words":1704,"readingMinutes":8,"sources":4,"figures":0,"topics":{"souverainete":3,"marches":12,"methode":10},"faq":[{"question":"What is a company's AI maturity?","answer":"AI maturity means an organisation's ability to turn artificial-intelligence uses into managed, repeatable and controlled results, although an analysis based on public sources can observe only part of that capability."},{"question":"How is AI maturity measured here?","answer":"The July 2026 internal study applies an eight-dimension weighted grid and a 1-to-5 scale exclusively to public documents, making it an indicator of documented maturity rather than an internal audit."},{"question":"Does a large market capitalisation imply AI maturity?","answer":"The comparison cannot derive AI maturity from market capitalisation because the measures concern different objects and the twenty-company sample is too narrow to establish a general relationship."},{"question":"Which dimension appears least documented?","answer":"Data receives the lowest average in both internal score sheets, a result that mainly signals limited public evidence on data quality, traceability and governance."},{"question":"Which groups of companies were compared?","answer":"The CAC 40 and Nasdaq panels must be read with their respective selection dates because a list of ten large capitalisations changes with prices, share issuance and index composition."},{"question":"Is this ranking investment advice?","answer":"No, because a public-source AI-maturity score measures neither a security's valuation nor its prospective return nor its suitability for an individual's circumstances."}],"url":"https://paulantoinetual.fr/en/blog/cac-40-vs-nasdaq-top-10","markdownUrl":"https://paulantoinetual.fr/en/blog/cac-40-vs-nasdaq-top-10.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/cac-40-vs-nasdaq-top-10.a2ui.json","image":"https://paulantoinetual.fr/img/og/cac-40-vs-nasdaq-top-10-en.jpg"},{"slug":"ia-bourse-chaine-de-valeur-circuit-ferme","title":"Where AI value accumulates: margins, cross-financing and competition","description":"An economic reading of the AI value chain in 2026: reported revenue, announced investment, financial dependencies and industrial alternatives.","category":"Markets & AI","categoryKey":"marches","published":"2026-07-20","updated":"2026-09-06","tags":["AI","stock market","value chain","circular financing","AI bubble","Nvidia","capex","semiconductors","vendor financing","where value accrues","financial literacy","inference"],"words":1615,"readingMinutes":7,"sources":12,"figures":0,"topics":{"couts":3,"marches":9},"faq":[{"question":"Is AI a bubble in 2026?","answer":"The word ‘bubble’ compresses a situation in which a technology can create operating value while some valuations already embed extremely demanding growth scenarios."},{"question":"What is AI cross-financing?","answer":"It describes structures in which investors, suppliers and customers occupy several economic roles, potentially strengthening demand while making its quality harder to assess."},{"question":"How useful is the comparison with telecoms in 2000?","answer":"The telecoms precedent shows that useful infrastructure can be overfinanced and that vendor credit can weaken demand, without predicting the same outcome for AI."},{"question":"Can Nvidia be replaced?","answer":"Nvidia remains difficult to bypass for many large-scale training workloads, while inference and specialised workloads leave more room for AMD accelerators and chips designed by cloud providers."},{"question":"Is this article investment advice?","answer":"This article offers an educational framework for separating reported results, announced commitments, theses and scenarios, without recommending the purchase or sale of a security."}],"url":"https://paulantoinetual.fr/en/blog/ia-bourse-chaine-de-valeur-circuit-ferme","markdownUrl":"https://paulantoinetual.fr/en/blog/ia-bourse-chaine-de-valeur-circuit-ferme.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/ia-bourse-chaine-de-valeur-circuit-ferme.a2ui.json","image":"https://paulantoinetual.fr/img/og/ia-bourse-chaine-de-valeur-circuit-ferme-en.jpg"},{"slug":"retirer-git-pilotage-par-roadmap","title":"Removing Git from the active workflow without losing rollback: our roadmap-driven approach","description":"A Junyr field report: agents no longer access Git, while the earlier history, deployment images, backups and a decision registry each retain a distinct role.","category":"Junyr Method™","categoryKey":"methode","published":"2026-07-15","updated":"2026-09-06","tags":["Git","roadmap-driven development","spec-driven development","AI agents","agentic engineering","UltraCoding","human-in-the-loop","AI governance","version control","Junyr Method"],"words":1941,"readingMinutes":9,"sources":6,"figures":0,"topics":{"agents":10,"methode":11,"outils":10},"faq":[{"question":"Should you remove Git from your projects?","answer":"For most teams, no: our choice remains a special case in which writing is serialised, the former repository is archived and control functions are divided among several explicit mechanisms."},{"question":"What happens to the code history without Git in the active workflow?","answer":"It does not disappear on the changeover date, but its granularity changes thereafter: the earlier repository remains archived, while subsequent changes are documented through plans, journals and code snapshots."},{"question":"How do you roll back without git revert?","answer":"Production rollback uses deployment images, while source recovery and regression analysis rely on separate mechanisms that are not full equivalents of `git revert` or `git bisect`."},{"question":"Is roadmap-driven development the same as spec-driven development?","answer":"It is a narrower variant: spec-driven development makes the specification a reference for the code, whereas our registry mainly fixes the decisions required to authorise and verify a change."},{"question":"What happens if two agents want to edit the same file?","answer":"Implementation is serialised through a single writing right: several agents may investigate or review plans concurrently, but only one orchestration session applies changes to the code."},{"question":"What is a decision-complete plan?","answer":"It is a plan that an implementer can execute without reopening structural choices, while leaving mechanical details until the code is written."}],"url":"https://paulantoinetual.fr/en/blog/retirer-git-pilotage-par-roadmap","markdownUrl":"https://paulantoinetual.fr/en/blog/retirer-git-pilotage-par-roadmap.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/retirer-git-pilotage-par-roadmap.a2ui.json","image":"https://paulantoinetual.fr/img/og/retirer-git-pilotage-par-roadmap-en.jpg"},{"slug":"optimisation-contexte-llm-cache-kv-prompt-caching","title":"LLM context: sizing the window, KV cache and prompt caching","description":"A technical guide to selecting useful context, calculating KV memory and comparing API caches without confusing compute, latency and billing.","category":"AI Agents","categoryKey":"agents","published":"2026-07-12","updated":"2026-09-06","tags":["LLM","context window","KV cache","prompt caching","RAG","context engineering","FinOps","Junyr Method"],"words":1750,"readingMinutes":8,"sources":15,"figures":0,"topics":{"couts":3,"methode":3,"prompts":10,"agents":3},"faq":[{"question":"What is context dilution in an LLM?","answer":"Context dilution is a loss of answer quality caused by irrelevant, competing or poorly positioned information, even when the prompt contains the evidence the model needs."},{"question":"Why can information in the middle of a long context be harder to use?","answer":"The Lost in the Middle study found on its retrieval tasks that several models often used information at the beginning or end more successfully than the same information in the middle."},{"question":"How do you calculate KV cache size?","answer":"For a dense transformer without special compression, KV cache grows linearly with layers, KV heads, head dimension, sequence length, batch size and bytes per element."},{"question":"Does prompt caching work the same way across providers?","answer":"Prompt caching reuses a previously processed prefix, but activation, thresholds, retention, write or storage charges and read prices depend on the model, API and date."},{"question":"Should you fill the entire context window?","answer":"The right length is the smallest window that preserves the necessary material with a measured margin, because more tokens always consume resources without guaranteeing a better answer."}],"url":"https://paulantoinetual.fr/en/blog/optimisation-contexte-llm-cache-kv-prompt-caching","markdownUrl":"https://paulantoinetual.fr/en/blog/optimisation-contexte-llm-cache-kv-prompt-caching.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/optimisation-contexte-llm-cache-kv-prompt-caching.a2ui.json","image":"https://paulantoinetual.fr/img/og/optimisation-contexte-llm-cache-kv-prompt-caching-en.jpg"},{"slug":"ingenierie-systemes-agentiques-human-in-the-loop","title":"Engineering agentic systems: golden rules, architecture and security of the “Human-in-the-Loop”","description":"HITL, HOTL and HOOTL: choosing the right supervision regime, enforcing executable limits and calibrating human escalation without mistaking statistical coverage for certainty.","category":"AI Agents","categoryKey":"agents","published":"2026-07-04","updated":"2026-09-06","tags":["AI agents","Human-in-the-Loop","HITL","AI governance","agentic engineering","12-Factor Agents","Model Context Protocol","A2A","agent security","conformal prediction","AgentOps","digital sovereignty","Junyr Method"],"words":2677,"readingMinutes":12,"sources":11,"figures":0,"topics":{"agents":12,"souverainete":9,"methode":6,"prompts":3,"outils":3},"faq":[{"question":"What is the difference between HITL, HOTL and HOOTL?","answer":"HITL, HOTL and HOOTL allocate authority differently between an agent and its supervisor according to risk, reversibility and the practical ability to interrupt the action."},{"question":"Should an agent's safeguards be coded into its prompt?","answer":"Prompt instructions can guide the model, but non-negotiable limits must be enforced by independent software controls where an action is authorised or executed."},{"question":"How can human validators be prevented from being overwhelmed with approval requests?","answer":"An escalation policy combines business criticality with calibrated uncertainty so that human attention is reserved for ambiguous or risky decisions without treating a conformal output as certain."},{"question":"What are the “12-Factor Agents”?","answer":"The 12-Factor Agents are a practical design guide that adapts principles for controlling context, tools, state and execution flow to agentic applications."},{"question":"Which level of supervision should an agent deployment begin with?","answer":"A new use case with material effects should generally begin with close human validation, then gain autonomy when production evidence and tested fallback mechanisms support the change."}],"url":"https://paulantoinetual.fr/en/blog/ingenierie-systemes-agentiques-human-in-the-loop","markdownUrl":"https://paulantoinetual.fr/en/blog/ingenierie-systemes-agentiques-human-in-the-loop.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/ingenierie-systemes-agentiques-human-in-the-loop.a2ui.json","image":"https://paulantoinetual.fr/img/og/ingenierie-systemes-agentiques-human-in-the-loop-en.jpg"},{"slug":"irp-cse-transformation-ia-jurisprudence","title":"CSE consultation and AI transformation: the right time and scope","description":"Three interim orders clarify CSE consultation for AI projects in France: effects on work, pilots, replacement tools, expert support and the separate AI Act framework.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-30","updated":"2026-09-06","tags":["CSE","IRP","social dialogue","Code du travail","consultation","case law","human resources","Junyr Method","AI governance","AgentOps"],"words":1581,"readingMinutes":7,"sources":12,"figures":0,"topics":{"agents":3,"methode":12},"faq":[{"question":"Must the CSE be consulted before every corporate AI project?","answer":"Not automatically: in a company subject to the consultative powers that apply from 50 employees, the employer must assess the decision and its effects under Article L.2312-8 of the French Code du travail."},{"question":"What does a company risk if it implements the project too early?","answer":"The orders examined show that an interim-relief judge may suspend use or deployment until consultation has ended and may attach a penalty payment to that measure."},{"question":"Does a software update with an AI feature trigger consultation?","answer":"It may do so when its functions, scope or effects change work sufficiently, as illustrated by the TJ Nanterre order of 29 January 2026."},{"question":"How should CSE consultation and the AI Act be combined?","answer":"The French Code du travail and the AI Act should be addressed together within the project while retaining their separate triggers, timetables and addressees."},{"question":"What resources can the CSE use to understand the project?","answer":"The CSE must receive precise information enabling it to assess the project’s scope and may, in the cases specified by the Code du travail, appoint an accredited expert."}],"url":"https://paulantoinetual.fr/en/blog/irp-cse-transformation-ia-jurisprudence","markdownUrl":"https://paulantoinetual.fr/en/blog/irp-cse-transformation-ia-jurisprudence.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/irp-cse-transformation-ia-jurisprudence.a2ui.json","image":"https://paulantoinetual.fr/img/og/irp-cse-transformation-ia-jurisprudence-en.jpg"},{"slug":"la-fin-des-emails-et-suites-office-2030","title":"Email and office suites: taking control of a fragmented workspace","description":"Email and office suites are not disappearing: organisations can reduce fragmentation by giving each channel a clear role and requiring interoperability, modularity and reversibility.","category":"Transition Management","categoryKey":"transition","published":"2026-06-26","updated":"2026-09-06","tags":["digital sovereignty","information overload","office suites","Microsoft 365","SecNumCloud","France 2030","CSCW","future of work","Junyr Method™","email","GDPR"],"words":1759,"readingMinutes":8,"sources":15,"figures":0,"topics":{"souverainete":6,"methode":8,"productivite":7,"outils":3},"faq":[{"question":"Will email really disappear by 2030?","answer":"No available evidence establishes an expiry date for email, although several uses are better moved out of the inbox."},{"question":"What is a governed workspace?","answer":"It is a set of channels coordinated by common rules, without requiring one supplier to provide every component."},{"question":"Why should an SME consider sovereign solutions?","answer":"Sovereignty becomes a practical criterion when jurisdiction, data sensitivity or supplier dependency changes business risk."},{"question":"Can a unified surface create another lock-in?","answer":"It can when integration makes data, identities or workflows difficult to move."},{"question":"Where should an organisation start?","answer":"Begin with a short channel inventory and choose one painful workflow whose result can be measured."},{"question":"What does the internal analysis of 874 conversations show?","answer":"It illustrates substitution opportunities in one micro-business, rather than measuring a rate that applies to SMEs generally."}],"url":"https://paulantoinetual.fr/en/blog/la-fin-des-emails-et-suites-office-2030","markdownUrl":"https://paulantoinetual.fr/en/blog/la-fin-des-emails-et-suites-office-2030.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/la-fin-des-emails-et-suites-office-2030.a2ui.json","image":"https://paulantoinetual.fr/img/og/la-fin-des-emails-et-suites-office-2030-en.jpg"},{"slug":"editeur-logiciels-ia-feature-moat","title":"SaaS vendors: building an advantage beyond the AI feature","description":"An AI feature alone is quickly copied; durable advantage comes from workflow integration, reliability, data, distribution and controlled unit economics.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-23","updated":"2026-09-06","tags":["SaaS","software vendor","technical debt","modernisation","AgentOps","orchestration","AI Act","Junyr Method"],"words":1859,"readingMinutes":8,"sources":6,"figures":0,"topics":{"agents":7,"souverainete":3,"marches":6,"methode":5,"outils":3},"faq":[{"question":"Can coding agents clear technical debt?","answer":"They can accelerate a well-scoped modernisation programme, but their speed replaces neither domain knowledge, testing nor human review."},{"question":"Will AI make software developers disappear?","answer":"Automation shifts some work towards specifying intent, architecture, evaluation and delivery accountability, without establishing that every role will follow the same path."},{"question":"Why does faster coding not guarantee more value?","answer":"Local throughput creates value only when the rest of the delivery system can absorb changes without increasing defects, rework or operational risk."},{"question":"When is a software vendor a provider under the AI Act?","answer":"The role depends on the system developed or commissioned, how it is placed on the market or put into service, and whose name it bears, rather than on the mere presence of an AI feature."}],"url":"https://paulantoinetual.fr/en/blog/editeur-logiciels-ia-feature-moat","markdownUrl":"https://paulantoinetual.fr/en/blog/editeur-logiciels-ia-feature-moat.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/editeur-logiciels-ia-feature-moat.a2ui.json","image":"https://paulantoinetual.fr/img/og/editeur-logiciels-ia-feature-moat-en.jpg"},{"slug":"objet-publicitaire-ia-desintermediation","title":"AI and promotional products: defend the client relationship through data and advice","description":"A practical framework for testing four AI uses in promotional products, protecting the client relationship and substantiating environmental claims.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-23","updated":"2026-09-06","tags":["promotional products","media object","AI","disintermediation","CSR","AI Act","Junyr Method","SME"],"words":1195,"readingMinutes":5,"sources":7,"figures":0,"topics":{"souverainete":3,"methode":5},"faq":[{"question":"Can AI support growth for a promotional-products distributor?","answer":"AI may improve margin, responsiveness or retention in particular workflows, but none of the sector data cited here supports a promise of automatic growth."},{"question":"What disintermediation risk does AI create for a distributor?","answer":"The risk arises when an assistant or platform controls product discovery while the distributor is confined to fulfilling the order."},{"question":"Is an AI-generated product mock-up a deepfake under the AI Act?","answer":"The classification depends on the visual and its context because the AI Act covers content that appreciably resembles an existing person, object, place, entity or event and would falsely appear authentic or truthful."},{"question":"Where should an AI transformation in promotional-products distribution begin?","answer":"Begin with a narrow workflow whose data, errors and commercial outcome can be measured before automating more of it."}],"url":"https://paulantoinetual.fr/en/blog/objet-publicitaire-ia-desintermediation","markdownUrl":"https://paulantoinetual.fr/en/blog/objet-publicitaire-ia-desintermediation.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/objet-publicitaire-ia-desintermediation.a2ui.json","image":"https://paulantoinetual.fr/img/og/objet-publicitaire-ia-desintermediation-en.jpg"},{"slug":"travail-asynchrone-tout-le-monde-devient-chef","title":"AI agents and asynchronous work: delegating without confusing supervision with management","description":"A practical framework for assigning work to AI agents in deferred time: mandates, permissions, control points, accountability and measurement, without turning every employee into a manager.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-23","updated":"2026-09-06","tags":["Junyr Method™","AI Agents","SME AI Transformation","Management"],"words":1297,"readingMinutes":6,"sources":5,"figures":0,"topics":{"agents":7,"methode":11,"productivite":3},"faq":[{"question":"Why do we describe some work with AI agents as asynchronous?","answer":"Work becomes asynchronous when a person assigns a task to an agent, continues with another activity, then returns to the result at an agreed point."},{"question":"Does delegating to an agent make every employee a manager?","answer":"Delegating to software uses some supervisory skills, but it confers no management title, hierarchical authority or new position under an employment contract."},{"question":"Is asynchronous work always more productive?","answer":"Asynchronous execution can free attention when tasks are separable and controllable, but its value depends on the cost of coordination, review and rework."},{"question":"Who is accountable for work produced by an AI agent?","answer":"The organisation should name the people who authorise the mandate, approve the output, administer the system and own decisions made with its results."},{"question":"Where should an SME begin?","answer":"An SME can begin with a narrow, reversible and measurable process, then decide whether to expand it from observed results rather than a general promise."},{"question":"How can teams limit deskilling?","answer":"Teams can preserve judgement by retaining learning tasks, explaining review decisions and regularly practising selected work without assistance."}],"url":"https://paulantoinetual.fr/en/blog/travail-asynchrone-tout-le-monde-devient-chef","markdownUrl":"https://paulantoinetual.fr/en/blog/travail-asynchrone-tout-le-monde-devient-chef.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/travail-asynchrone-tout-le-monde-devient-chef.a2ui.json","image":"https://paulantoinetual.fr/img/og/travail-asynchrone-tout-le-monde-devient-chef-en.jpg"},{"slug":"vibe-coding-mort-ultracoding","title":"Beyond vibe coding: four practices for verifiable software","description":"Vibe coding remains useful for exploration; in production, UltraCoding adds orchestration, critical review, three-layer testing and explicit release gates.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-23","updated":"2026-09-06","tags":["vibe coding","UltraCoding","agentic engineering","multi-agent","adversarial review","software quality","AI testing","code security","sovereignty","AI FinOps","Junyr Method"],"words":1549,"readingMinutes":7,"sources":12,"figures":0,"topics":{"agents":7,"couts":3,"souverainete":6,"methode":5,"outils":10},"faq":[{"question":"Is vibe coding really dead?","answer":"Vibe coding remains useful for disposable exploration, but its principle of leaving generated code unread is not a delivery method for software that must be secured, maintained and operated."},{"question":"What is UltraCoding?","answer":"UltraCoding is the name used here for an agent-assisted development method that organises generation, challenge, technical controls and human judgement around explicit release criteria."},{"question":"Does Veracode's benchmark mean my code has a 45% chance of being vulnerable?","answer":"No, 45% is the proportion of samples that failed security tests in Veracode's 2025 benchmark, not a predictive probability that applies to every production system."},{"question":"Do multiple agents always deliver a better result?","answer":"No, gains depend on the type of work, how far it can be parallelised and the quality of orchestration, while strong dependencies can make multiple agents costlier and harder to coordinate."},{"question":"Does an agent reviewer provide independent assurance?","answer":"A critic agent adds another perspective and may detect faults, but it does not provide independent assurance by itself when it shares the generator's provider, assumptions or environment."},{"question":"Can AI judge interface quality on its own?","answer":"No, a visual model can detect rendering anomalies and prepare a review, while understanding users, real accessibility and design trade-offs remain human responsibilities."}],"url":"https://paulantoinetual.fr/en/blog/vibe-coding-mort-ultracoding","markdownUrl":"https://paulantoinetual.fr/en/blog/vibe-coding-mort-ultracoding.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/vibe-coding-mort-ultracoding.a2ui.json","image":"https://paulantoinetual.fr/img/og/vibe-coding-mort-ultracoding-en.jpg"},{"slug":"junyr-vous-nachetez-pas-une-ia-vous-achetez-la-base","title":"Junyr: a business foundation connected to your AI assistants through MCP","description":"Junyr separates the business foundation, operational functions and AI model to make architecture, data control and running costs easier to assess.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-14","updated":"2026-09-06","tags":["MCP","digital sovereignty","FinOps","AI agents","AI","SME","tokens","Junyr Method"],"words":1494,"readingMinutes":7,"sources":7,"figures":0,"topics":{"agents":9,"couts":7,"souverainete":3,"methode":5,"outils":3},"faq":[{"question":"What is the central proposition of Junyr Suite?","answer":"Junyr Suite combines a business foundation and management functions with an MCP interface so that compatible assistants can consult authorised data and trigger the actions exposed by the platform."},{"question":"Why separate Junyr's price from the cost of the AI model?","answer":"This separation distinguishes the platform fee from the costs of the chosen model, clarifying each cost item without guaranteeing that total expenditure will be fixed or lower in every configuration."},{"question":"Does a Claude, ChatGPT or Gemini subscription always replace an API?","answer":"A subscription and an API are separate products, and the right choice depends on available connectors, provider terms, usage limits and the degree of automation required."},{"question":"What does data sovereignty mean in this architecture?","answer":"Sovereignty rests on verifiable commitments covering hosting, access, transfers, export and deletion, while using a third-party model adds a separate processing perimeter."},{"question":"How does MCP support interoperability?","answer":"MCP is an open standard through which a server can expose tools, resources and prompts to compatible AI applications by means of a common interface."},{"question":"How should the cost of a subscription and an API be compared?","answer":"A sound comparison applies current model prices to observed volumes and then adds every other cost over the same period and for the same service delivered."},{"question":"How do you get started with Junyr?","answer":"The published entry point is a free, no-commitment 30-minute AI maturity audit by video call, before any decision about consulting support or Junyr Suite."}],"url":"https://paulantoinetual.fr/en/blog/junyr-vous-nachetez-pas-une-ia-vous-achetez-la-base","markdownUrl":"https://paulantoinetual.fr/en/blog/junyr-vous-nachetez-pas-une-ia-vous-achetez-la-base.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/junyr-vous-nachetez-pas-une-ia-vous-achetez-la-base.a2ui.json","image":"https://paulantoinetual.fr/img/og/junyr-vous-nachetez-pas-une-ia-vous-achetez-la-base-en.jpg"},{"slug":"ou-va-la-valeur-de-lia-trois-signaux-de","title":"Where is AI value created? Three signals for deciding without betting on one supplier","description":"Open models, local execution and reproducible software: a careful method for investing in data, integration and skills.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-07","updated":"2026-09-06","tags":["value chain","commoditisation","business leaders","AI","Junyr Method","open-weight","SME","sovereignty"],"words":1297,"readingMinutes":6,"sources":11,"figures":0,"topics":{"souverainete":3,"methode":6},"faq":[{"question":"Is an open-weight model as good as a proprietary one?","answer":"The licence does not predict quality: the comparison must cover a specific version, configuration and set of real business tasks."},{"question":"Should an SME run its models locally?","answer":"Local execution is credible for some workflows, but its value depends on the model, volume, latency, data and available skills."},{"question":"Does a lower price per token necessarily reduce the bill?","answer":"A cheaper token lowers one unit price, while the total bill also depends on volume, rework and human review."},{"question":"Is an AI agent's software a durable barrier?","answer":"Orchestration code can be reproduced or replaced, whereas operational reliability and workflow access require continuing work."},{"question":"Where should an AI budget be concentrated?","answer":"The most resilient budget funds assets that remain useful when the model or tool changes."}],"url":"https://paulantoinetual.fr/en/blog/ou-va-la-valeur-de-lia-trois-signaux-de","markdownUrl":"https://paulantoinetual.fr/en/blog/ou-va-la-valeur-de-lia-trois-signaux-de.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/ou-va-la-valeur-de-lia-trois-signaux-de.a2ui.json","image":"https://paulantoinetual.fr/img/og/ou-va-la-valeur-de-lia-trois-signaux-de-en.jpg"},{"slug":"golden-standard-prospection-visio-2026","title":"Video prospecting: running a useful first B2B meeting in 2026","description":"Prepare, discover, demonstrate, quantify and close a first remote B2B meeting, with recording and AI kept within a clear governance framework.","category":"Personal Effectiveness","categoryKey":"efficacite","published":"2026-06-02","updated":"2026-09-06","tags":["B2B","business leaders","AI","Junyr Method","prospecting","sales transformation","sales","video conferencing"],"words":1752,"readingMinutes":8,"sources":19,"figures":0,"topics":{"methode":7,"productivite":9},"faq":[{"question":"How long should a first video prospecting meeting be?","answer":"A thirty-to-forty-five-minute slot is a workable starting point, to be adjusted for the complexity of the subject, the number of participants and the decision expected."},{"question":"What talk-to-listen ratio should a sales meeting aim for?","answer":"The 43% talking to 57% listening ratio observed by Gong in its sample can inform a debrief, without becoming a universal rule or a second-by-second target [8]."},{"question":"Can a meeting be recorded and transcribed in line with the GDPR?","answer":"Yes in some settings, but a serious policy must distinguish the lawful basis for processing, the information given to people, their rights and the criminal-law constraints that protect private or confidential speech [17, 18]."},{"question":"Is an AI meeting-summary assistant useful?","answer":"It is useful when it shortens the path from meeting to verified summary, provided the team measures total time, corrections and errors instead of assuming an automatic gain."},{"question":"Does self-service make the B2B sales representative disappear?","answer":"Gartner's 2026 surveys describe two coexisting expectations: 67% of respondents prefer a rep-free experience for part of the journey, while 69% seek a sales representative to validate AI-generated insights [5, 6]."},{"question":"How do you move an opportunity forward after the first meeting?","answer":"Progress becomes reviewable when buyer and seller leave with the same summary, an explicit next decision and dated responsibilities."}],"url":"https://paulantoinetual.fr/en/blog/golden-standard-prospection-visio-2026","markdownUrl":"https://paulantoinetual.fr/en/blog/golden-standard-prospection-visio-2026.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/golden-standard-prospection-visio-2026.a2ui.json","image":"https://paulantoinetual.fr/img/og/golden-standard-prospection-visio-2026-en.jpg"},{"slug":"junyr-agents-deleguer-lia-dans-votre","title":"Junyr Agents™: delegating business processes to AI with a clear mandate","description":"A practical method for governing an AI agent: a versioned mandate, permissions, human approvals, an action log and a measured pilot.","category":"Junyr Method™","categoryKey":"methode","published":"2026-06-01","updated":"2026-09-06","tags":["AI agents","AI Act","governance","AI","Junyr Method","SME"],"words":1440,"readingMinutes":7,"sources":4,"figures":0,"topics":{"agents":7,"souverainete":3,"methode":9},"faq":[{"question":"What is the difference between an AI assistant and an AI agent?","answer":"The operational difference lies in the degree of delegated action: an assistant produces an answer for review, whereas an agent can complete a sequence of steps and call tools within an authorised scope."},{"question":"Does the AI Act require every AI agent to keep a log?","answer":"No: obligations depend in particular on the system’s classification, its use and the company’s role, although a comprehensive log remains sound governance for any agent that can take action."},{"question":"How should an SME launch its first AI agent?","answer":"The first pilot should cover a frequent, measurable and reversible sub-process, with a baseline and stop conditions agreed before the first trial."},{"question":"What does the Junyr AI maturity audit include?","answer":"The Junyr audit is a free, no-commitment 30-minute video call that provides an actionable initial diagnosis rather than promising a complete transformation."}],"url":"https://paulantoinetual.fr/en/blog/junyr-agents-deleguer-lia-dans-votre","markdownUrl":"https://paulantoinetual.fr/en/blog/junyr-agents-deleguer-lia-dans-votre.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/junyr-agents-deleguer-lia-dans-votre.a2ui.json","image":"https://paulantoinetual.fr/img/og/junyr-agents-deleguer-lia-dans-votre-en.jpg"},{"slug":"la-fin-du-telephone-pourquoi-le-binome","title":"From unsolicited calls to messaging and video: organising customer contact in 2026","description":"Voice still matters, but unannounced calls are losing effectiveness: a practical method for assigning clear roles to messaging, video and controlled automation.","category":"Transition Management","categoryKey":"transition","published":"2026-05-31","updated":"2026-09-06","tags":["telephone","video conferencing","messaging","AI","sales transformation","SME","RCS"],"words":1340,"readingMinutes":6,"sources":16,"figures":0,"topics":{"methode":6,"productivite":3},"faq":[{"question":"Will the telephone disappear from business?","answer":"Voice retains several decisive uses, while an unannounced call is becoming a less reliable entry point for customer contact."},{"question":"What changed for telephone canvassing on 11 August 2026?","answer":"The French Consumer Code now prohibits telephone marketing to a consumer without prior consent, subject to the exceptions set out in the legislation."},{"question":"Why combine messaging and video?","answer":"The two channels cover different moments in one journey and avoid imposing a synchronous conversation on every interaction."},{"question":"Can an AI appointment assistant comply with the GDPR?","answer":"Compliance depends on the specific processing, its lawful basis and its safeguards; consent is not the GDPR's only lawful basis."},{"question":"Should an organisation leave Teams or Google Meet?","answer":"The decision depends on the data, applicable jurisdictions, supplier dependency and total cost rather than the national origin of a brand in isolation."}],"url":"https://paulantoinetual.fr/en/blog/la-fin-du-telephone-pourquoi-le-binome","markdownUrl":"https://paulantoinetual.fr/en/blog/la-fin-du-telephone-pourquoi-le-binome.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/la-fin-du-telephone-pourquoi-le-binome.a2ui.json","image":"https://paulantoinetual.fr/img/og/la-fin-du-telephone-pourquoi-le-binome-en.jpg"},{"slug":"cryptographie-2026-transport-stockage","title":"Cryptography 2026: transport, storage, identities and the post-quantum transition, the decision framework for business leaders and CISOs","description":"Cryptography has become a leadership decision. Data in transit, at rest, identities and post-quantum: the 2026 framework for business leaders.","category":"Junyr Method™","categoryKey":"methode","published":"2026-05-29","updated":"2026-09-06","tags":["ANSSI","encryption","cryptography","FIDO2","AI","Junyr Method","NIST","passkeys","SME","post-quantum","digital sovereignty"],"words":2764,"readingMinutes":13,"sources":43,"figures":0,"topics":{"souverainete":10,"methode":8,"outils":3},"faq":[{"question":"When should an SME begin its post-quantum transition?","answer":"The right trigger is where the required confidentiality lifetime of the data, the time needed to migrate and uncertainty about an exploitable quantum threat intersect."},{"question":"Does the HSM decrypt all data in envelope encryption?","answer":"In the usual architecture, the KMS or HSM protects the master key and decrypts the data key, while the authorised application decrypts the bulk data with that key."},{"question":"Does a hash or MAC make a backup immutable?","answer":"A hash detects a change and a MAC authenticates its source, but neither prevents deletion, replacement or rollback to an older version on its own."},{"question":"Does quantum computing already make AES-128 vulnerable?","answer":"No: NIST considers that AES-128 should remain secure for decades despite Grover's theoretical advantage, while AES-256 provides additional margin for new sensitive architectures."},{"question":"Do passkeys block every form of phishing?","answer":"WebAuthn resists credential theft through a false domain by binding authentication to the origin, but account security still depends on the endpoint, recovery, fallback methods and session protection."},{"question":"Does eIDAS 2.0 require browsers to integrate every authority on European trusted lists?","answer":"The Commission says that browsers must recognise QWACs and display their identity data, but are not required to integrate or directly use national trusted lists."}],"url":"https://paulantoinetual.fr/en/blog/cryptographie-2026-transport-stockage","markdownUrl":"https://paulantoinetual.fr/en/blog/cryptographie-2026-transport-stockage.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/cryptographie-2026-transport-stockage.a2ui.json","image":"https://paulantoinetual.fr/img/og/cryptographie-2026-transport-stockage-en.jpg"},{"slug":"echelle-methode-junyr-cinq-niveaux-maturite-ia-pme","title":"The Junyr Method™ Scale: assess an SME’s 5 AI maturity levels","description":"Spectateur, Artisan, Orchestre, Architecte, Pionnier: assess your SME’s AI maturity across nine dimensions using evidence and cumulative prerequisites.","category":"Junyr Method™","categoryKey":"methode","published":"2026-05-23","updated":"2026-09-06","tags":[],"words":2142,"readingMinutes":10,"sources":0,"figures":0,"topics":{"methode":6},"faq":[],"url":"https://paulantoinetual.fr/en/blog/echelle-methode-junyr-cinq-niveaux-maturite-ia-pme","a2uiUrl":"https://paulantoinetual.fr/en/blog/echelle-methode-junyr-cinq-niveaux-maturite-ia-pme.a2ui.json","image":"https://paulantoinetual.fr/img/og/echelle-methode-junyr-cinq-niveaux-maturite-ia-pme-en.jpg"},{"slug":"la-fin-du-prompt-engineering-pourquoi","title":"Beyond prompt engineering: designing reliable context for AI","description":"Move from isolated requests to tested instructions: context, examples, data, tools and output formats for dependable business AI.","category":"Transition Management","categoryKey":"transition","published":"2026-05-19","updated":"2026-09-06","tags":["context engineering","prompt engineering","AI","Junyr Method","SME","AI agents"],"words":1318,"readingMinutes":6,"sources":8,"figures":0,"topics":{"agents":3,"methode":5,"prompts":10},"faq":[{"question":"Is prompt engineering really dead?","answer":"The term remains useful, but a dependable application requires teams to design everything the model receives and everything that controls its actions, beyond the wording of one request."},{"question":"Should every prompt use XML?","answer":"XML can separate the parts of a complex context, while clear prose, Markdown headings or a native schema may suit other tasks better."},{"question":"What is the difference between prompt engineering and context engineering?","answer":"Prompt engineering develops the instruction, whereas context engineering organises all the information and capabilities available when the model must answer or act."},{"question":"Does a non-technical team need to learn XML?","answer":"A business team mainly needs to define a testable task, while a designated owner can translate that definition into the format suited to the model and application."},{"question":"Does a structured instruction prevent prompt injection?","answer":"Separating instructions from data reduces some ambiguity, but it is only one layer of a defence that must also limit permissions and review consequential actions."},{"question":"Does markup replace RAG or fine-tuning?","answer":"Markup organises a request, RAG selects references and fine-tuning changes model behaviour, so the three techniques address different problems."}],"url":"https://paulantoinetual.fr/en/blog/la-fin-du-prompt-engineering-pourquoi","markdownUrl":"https://paulantoinetual.fr/en/blog/la-fin-du-prompt-engineering-pourquoi.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/la-fin-du-prompt-engineering-pourquoi.a2ui.json","image":"https://paulantoinetual.fr/img/og/la-fin-du-prompt-engineering-pourquoi-en.jpg"},{"slug":"budgets-tokens-et-api-ia-le-guide","title":"Token and AI API budgets: the FinOps guide for SMEs in 2026","description":"A 2026 FinOps guide to measuring, budgeting and controlling AI API costs, from model routing to genuinely attributable ROI.","category":"Personal Effectiveness","categoryKey":"efficacite","published":"2026-05-18","updated":"2026-09-06","tags":["AI Act","FinOps","AI","ISO 42001","LLM Gateway","SME","tokens"],"words":3261,"readingMinutes":15,"sources":11,"figures":0,"topics":{"couts":10,"souverainete":3,"productivite":3},"faq":[],"url":"https://paulantoinetual.fr/en/blog/budgets-tokens-et-api-ia-le-guide","markdownUrl":"https://paulantoinetual.fr/en/blog/budgets-tokens-et-api-ia-le-guide.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/budgets-tokens-et-api-ia-le-guide.a2ui.json","image":"https://paulantoinetual.fr/img/og/budgets-tokens-et-api-ia-le-guide-en.jpg"},{"slug":"souverainete-numerique-les-3","title":"Cloud, local or hybrid in 2026: five sovereignty decisions for an SME","description":"A practical method for deciding on jurisdiction, source-code data, cybersecurity, total cost and post-quantum migration without treating cloud and on-premise as doctrines.","category":"Cybersecurity","categoryKey":"cyber","published":"2026-05-13","updated":"2026-09-06","tags":["ANSSI","CLOUD Act","cybersecurity","GTIG","AI","Microsoft 365","on-premise","cloud outages","SME","post-quantum","slopsquatting","SNDL","digital sovereignty","Splinternet","zero-day"],"words":1757,"readingMinutes":8,"sources":10,"figures":0,"topics":{"souverainete":12,"outils":3},"faq":[{"question":"Does the CLOUD Act give automatic access to all data hosted in Europe by a US company?","answer":"No: it requires certain providers subject to US jurisdiction to respond to valid legal process for data within their possession, custody or control, regardless of where that data is stored."},{"question":"Does GitHub Copilot use code from every plan to train models?","answer":"No: from 24 April 2026, training use applies by default to interactions on individual Free, Pro and Pro+ plans, with an opt-out, while Business and Enterprise are excluded from this change."},{"question":"Does a local LLM cost nothing?","answer":"No: removing a per-token bill does not remove hardware, electricity, operations or security costs, and the economics depend on actual volume and service requirements."},{"question":"Does ANSSI require every SME to adopt post-quantum cryptography in 2027?","answer":"No: the 2027 objective concerns products entering qualification, while regulatory duties depend on scope and ANSSI advises every organisation to begin an inventory now."},{"question":"Which hosting model should an SME choose?","answer":"The sound choice assigns each workload to public cloud, a qualified service, a private environment or a hybrid architecture according to its constraints rather than one universal doctrine."}],"url":"https://paulantoinetual.fr/en/blog/souverainete-numerique-les-3","markdownUrl":"https://paulantoinetual.fr/en/blog/souverainete-numerique-les-3.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/souverainete-numerique-les-3.a2ui.json","image":"https://paulantoinetual.fr/img/og/souverainete-numerique-les-3-en.jpg"},{"slug":"cinq-erreurs-fatales-transformation-ia-pme","title":"Five framing mistakes that weaken an SME AI transformation","description":"Tool-first choices, aimless PoCs, an isolated sponsor, poorly measured value and unprepared data: five observable risks and the decisions that control them.","category":"Methodology","categoryKey":"methodologie","published":"2026-04-27","updated":"2026-09-06","tags":[],"words":1375,"readingMinutes":6,"sources":0,"figures":0,"topics":{"methode":5},"faq":[],"url":"https://paulantoinetual.fr/en/blog/cinq-erreurs-fatales-transformation-ia-pme","a2uiUrl":"https://paulantoinetual.fr/en/blog/cinq-erreurs-fatales-transformation-ia-pme.a2ui.json","image":"https://paulantoinetual.fr/img/og/cinq-erreurs-fatales-transformation-ia-pme-en.jpg"},{"slug":"industec-roi-182-pourcent-9-mois","title":"INDUSTEC: measuring the ROI of an AI quotation assistant","description":"INDUSTEC is an illustrative example for scoping an AI quotation assistant, measuring time, quality and adoption, and calculating ROI while keeping released capacity separate from cash savings.","category":"Illustrative example","categoryKey":"exemple","published":"2026-04-27","updated":"2026-09-06","tags":[],"words":1936,"readingMinutes":9,"sources":0,"figures":0,"topics":{"couts":4},"faq":[],"url":"https://paulantoinetual.fr/en/blog/industec-roi-182-pourcent-9-mois","a2uiUrl":"https://paulantoinetual.fr/en/blog/industec-roi-182-pourcent-9-mois.a2ui.json","image":"https://paulantoinetual.fr/img/og/industec-roi-182-pourcent-9-mois-en.jpg"},{"slug":"du-gagnant-gagnant-au-hard-bargaining","title":"From win-win to considered firmness: negotiating when trust declines","description":"A practical framework for combining cooperation and bargaining power: a real BATNA, reservation point, justified anchor, verifiable consequences and a way back.","category":"Transition Management","categoryKey":"transition","published":"2025-11-05","updated":"2026-09-06","tags":[],"words":1080,"readingMinutes":5,"sources":0,"figures":0,"topics":{"methode":3},"faq":[],"url":"https://paulantoinetual.fr/en/blog/du-gagnant-gagnant-au-hard-bargaining","markdownUrl":"https://paulantoinetual.fr/en/blog/du-gagnant-gagnant-au-hard-bargaining.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/du-gagnant-gagnant-au-hard-bargaining.a2ui.json","image":"https://paulantoinetual.fr/img/og/du-gagnant-gagnant-au-hard-bargaining-en.jpg"},{"slug":"la-paix-au-service-de-la-performance","title":"Mediation and Conciliation: Resolving Conflict Without Overpromising","description":"A precise legal and managerial framework for deciding when to use mediation or conciliation, securing the process and measuring its effects.","category":"Transition Management","categoryKey":"transition","published":"2025-09-01","updated":"2026-09-06","tags":[],"words":1352,"readingMinutes":6,"sources":0,"figures":0,"topics":{"outils":3,"methode":3},"faq":[],"url":"https://paulantoinetual.fr/en/blog/la-paix-au-service-de-la-performance","markdownUrl":"https://paulantoinetual.fr/en/blog/la-paix-au-service-de-la-performance.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/la-paix-au-service-de-la-performance.a2ui.json","image":"https://paulantoinetual.fr/img/og/la-paix-au-service-de-la-performance-en.jpg"},{"slug":"lia-en-tpe-une-revolution-similaire","title":"AI in micro-businesses: what 1980s computing can really teach us","description":"A practical method for choosing a first AI use case, measuring its full cost, protecting client data and deciding whether to continue or stop.","category":"Transition Management","categoryKey":"transition","published":"2024-11-19","updated":"2026-09-06","tags":[],"words":1073,"readingMinutes":5,"sources":0,"figures":0,"topics":{"methode":3},"faq":[{"question":"Which AI use case should a micro-business start with?","answer":"The strongest candidate is a frequent, low-risk, reversible and easily checked task owned by a domain expert who can assess every result."},{"question":"How can a business tell whether AI actually saves time?","answer":"It should compare a representative batch completed without AI and then with AI, counting all human and technical work required to reach an accepted result."},{"question":"Can a micro-business let AI reply to clients on its own?","answer":"Automated replies should be considered only after a successful supervised trial, within a narrow scope and with a clear route for escalation."},{"question":"Which client data can be sent to an AI tool?","answer":"The answer depends on the purpose, the supplier contract and the applicable rules, so the necessary data and its journey must be documented before any trial."}],"url":"https://paulantoinetual.fr/en/blog/lia-en-tpe-une-revolution-similaire","markdownUrl":"https://paulantoinetual.fr/en/blog/lia-en-tpe-une-revolution-similaire.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/lia-en-tpe-une-revolution-similaire.a2ui.json","image":"https://paulantoinetual.fr/img/og/lia-en-tpe-une-revolution-similaire-en.jpg"},{"slug":"accompagner-la-promotion-interne-dun","title":"Supporting the internal promotion of a sales manager in a communications agency","description":"Promoting a salesperson into management without weakening the team or margin: mandate, portfolio, agency objectives and support over 90 days.","category":"Transition Management","categoryKey":"transition","published":"2024-06-16","updated":"2026-09-06","tags":[],"words":988,"readingMinutes":4,"sources":0,"figures":0,"topics":{"methode":3},"faq":[],"url":"https://paulantoinetual.fr/en/blog/accompagner-la-promotion-interne-dun","markdownUrl":"https://paulantoinetual.fr/en/blog/accompagner-la-promotion-interne-dun.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/accompagner-la-promotion-interne-dun.a2ui.json","image":"https://paulantoinetual.fr/img/og/accompagner-la-promotion-interne-dun-en.jpg"},{"slug":"implementer-lintelligence-artificielle","title":"Implementing artificial intelligence to grow a communications agency","description":"A practical method for choosing a process, measuring its baseline, preparing the data and deciding whether an AI pilot deserves to expand.","category":"Transition Management","categoryKey":"transition","published":"2024-06-09","updated":"2026-09-06","tags":[],"words":1397,"readingMinutes":6,"sources":0,"figures":0,"topics":{"methode":3},"faq":[],"url":"https://paulantoinetual.fr/en/blog/implementer-lintelligence-artificielle","markdownUrl":"https://paulantoinetual.fr/en/blog/implementer-lintelligence-artificielle.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/implementer-lintelligence-artificielle.a2ui.json","image":"https://paulantoinetual.fr/img/og/implementer-lintelligence-artificielle-en.jpg"},{"slug":"recrutement-de-la-generation-z","title":"Recruiting early-career talent without generational stereotypes","description":"A practical method for defining the role, making the process transparent, assessing comparable evidence and presenting a realistic employment proposition.","category":"Transition Management","categoryKey":"transition","published":"2024-06-02","updated":"2026-09-06","tags":[],"words":1314,"readingMinutes":6,"sources":0,"figures":0,"topics":{"methode":3},"faq":[],"url":"https://paulantoinetual.fr/en/blog/recrutement-de-la-generation-z","markdownUrl":"https://paulantoinetual.fr/en/blog/recrutement-de-la-generation-z.md","a2uiUrl":"https://paulantoinetual.fr/en/blog/recrutement-de-la-generation-z.a2ui.json","image":"https://paulantoinetual.fr/img/og/recrutement-de-la-generation-z-en.jpg"}]}