From POC to industrialisation: understanding the AI paradox
94% of French companies have launched AI initiatives. Only 49% have managed to industrialise them. Hidden between those two figures is the industrialisation paradox: a costly gap that slows transformation down. Julien Laugel, Managing Director Data at ekino, uses four videos to unpack the reasons behind this widespread failure and the practical keys to industrialising your AI projects successfully. Source: Digital Radar 2026, an ekino × CSA study of more than 500 French CDOs
WHY DO AI PROJECTS FAIL?
The 2 fatal mistakes in AI projects
Technology is never the problem. AI projects fail for two structural reasons that most organisations underestimate.
- Mistake #1: treating AI as a technology topic rather than a company-wide transformation.
- Mistake #2: failing to specify the problem properly before deploying the solution.
- The 5 whys method: how to avoid the hammer-and-nail trap.
AI: escaping operational efficiency myopia
Generative AI is not only there to speed up what you were already doing. Limiting it to efficiency gains means missing out on 80% of its potential.
- The quick-wins trap: why focusing solely on operational efficiency cuts you off from the real potential.
- Understanding, not just generating: using AI to analyse and synthesise your information assets.
- The “less glamorous” uses: why non-generative use cases often carry more value.
HOW TO SUCCEED AT INDUSTRIALISATION
Acceptability: the hidden reason for failure
You have the best AI system in the world. But if your teams reject it, it will never reach production. Acceptability is the factor nobody budgets for.
- Human-machine coupling: why you have to design the whole system, not just the AI
- The oracle trap: when a lack of explainability breeds rejection
- Transformation and acculturation: the irreducible time you have to accept
AI governance: avoiding Pandora’s box
20 unregulated POCs are 20 time bombs. Agentic AI gives your LLMs arms and legs. It also gives them a capacity for harm that has to be governed.
- Incremental governance: how to move forward on value creation AND on structure at the same time
- The trap of the two extremes: paralysing rigidity vs. wild, unregulated POCs
- Agentic AI and risk: why connecting AI to your enterprise systems requires a strict framework
Your AI project deserves better than a POC gathering dust.
You have identified the mistakes to avoid. You now know why 51% of AI projects never reach production.
ekino supports companies in industrialising their AI projects: from specifying the need to going into production, by way of governance and acceptability.
We built AVA, the multi-LLM platform deployed to 23,000 Havas employees. We know what going from POC to ROI means.
Articles
Un projet ?
Parlez-nous de vous, de vos projets, de vos ambitions, de vos enjeux, et construisons ensemble des solutions concrètes qui contribuent à votre transformation.