Read the online version
The full white paper below, structured in chapters with anchors, freely available. No form.
Read online →White paper no. 2 · June 2026
From PoC to Industrialisation: Leading AI Change Management in European SMEs
The question for 2026 is no longer “should we do AI?” but “how do we move from PoC to industrialisation?”. Only 46% of AI PoCs reach production: a wall that is 70% human, not technological.
This white paper sets out the method: the seven frictions of the “last mile”, the five levers of change management, AgentOps, and a 12-month roadmap.
What you will find inside
- The “last mile” wall: why only 46% of AI PoCs reach production, and why the obstacle is organisational, not technological.
- The seven frictions of the last mile: the map of what blocks the path from a successful pilot to deployment at scale, and how to clear each one.
- The five levers of change management: the BCG 10-20-70 rule applied to the SME and mid-cap (70% of success rests on people and processes).
- AgentOps: extending DevOps/MLOps to observe and govern AI agents in production (a control plane, guardrails, human validation).
- ROI and how to fund it: €30,000 to €80,000 over 12 months for an SME or mid-cap of 50 to 500 staff, with funding available by scheme (Bpifrance Diag Data IA 25%, IA Booster up to 80%, EDIH).
- The 12-month roadmap: quarter by quarter, from framing to governed scaling.
Key figures
46%
only of AI PoCs reach production: half stay stuck at the pilot stage.
Source: IDC × Lenovo, CIO Playbook 2026
10-20-70
10% technology, 20% data and infrastructure, 70% people and processes: the rule of AI success.
Source: BCG, AI Transformation Is a Workforce Transformation, 2026
20.0%
of EU27 enterprises were using AI in 2025: 17% of small companies against 55% of large ones.
Source: Eurostat, December 2025
70.9%
of European companies that considered AI without adopting it cite a lack of skills.
Source: Eurostat, December 2025
30%
only of organisations redesign their core processes around AI. 37% use it superficially.
Source: Deloitte, State of AI in the Enterprise 2026
47%
of workers already spend more time directing and supervising AI than doing the work themselves.
Source: BCG, AI at Work 2026 (3 June 2026)
74%
of companies have already had to pull a customer-facing AI agent out of production: proof that industrialisation must be governed, not merely deployed.
Source: Sinch, May 2026 (2,527 decision-makers, 10 countries)
Access the white paper
Two ways to access the white paper, whichever you prefer:
Receive the PDF by email
For an offline copy (EN or FR). Click below: your email client opens with a pre-filled message.
Request the PDF by email →Or download the PDF directly (11 pages, no email) →
GDPR-compliant. One-click unsubscribe. No data shared with any third party.
You can also write directly to contact@paulantoinetual.fr.
Frequently asked questions
- Why do only 46% of AI PoCs reach production?
- Because the obstacle is organisational, not technical: the BCG 10-20-70 rule attributes 70% of success to people and processes. The “last mile” (Harvard Business Review, March 2026) is a problem of organisational design. The 46% figure comes from the IDC-Lenovo CIO Playbook 2026 study (January 2026): half of all PoCs stay stuck at the pilot stage.
- What is AgentOps?
- The operational discipline that extends DevOps/MLOps to observe and govern AI agents in production: a control plane, guardrails, human validation and an auditable log. Junyr Agents™ applies the Plan → Execute → Verify/Test pattern.
- How much does industrialising AI cost in an SME?
- €30,000 to €80,000 over 12 months for an SME or mid-cap of 50 to 500 staff. On the funding side (January 2026 conditions): Bpifrance's Diag Data IA covers 25% of a fixed-price €10,000 diagnostic, the IA Booster funds up to 80% of the support work, and the European Digital Innovation Hubs typically cover 50% of service costs.
- How does it differ from the first white paper?
- “Maturity of AI in French SMEs” answers “where do I stand?”; “From PoC to Industrialisation” answers “how do I industrialise?”, in the European context and with AgentOps.