In March, 2026, AXA has dedicated its yearly #AXAWeekForGrowth initiative, a week of events to upskill employees around the world - to #AI. During one of the sessions, our CEO Thomas Buberl had a very insightful fireside chat with Eléonore Crespo, CEO of Pigment, on how to foster AI adoption in large organizations.

Beyond the great energy of the discussion, it highlighted very concrete challenges — and some fundamental mindset gaps, that could apply to any large organization.

The startup mindset gaps applied to AI.

Yes, AI adoption requires daily practice in almost everything we do.

It also requires access to the best technology — including the latest models and their paid versions. While we can argue it costs ~€20/month, to what extent can a company, whatever its industry and its level of threat from AI, expect employees to pay out of pocket to stay competitive?

Yes, this daily practice requires time.

If you work in a startup/scale-up, or hold a senior role in a large company, you're probably used to working at evening or even Sunday afternoons. These moments often become opportunities to experiment with AI. But can we really expect this level of effort from all teams?

No, there is no magic formula.

Balancing legitimate security constraints, proper process controls with easy access for all to the technology, has no silver bullet. But there are a few pragmatic actions everyone can take.

My 5 recommendations to foster AI adoption within your organization.

1. Clarify the grey zones

What can employees safely do with LLMs on personal devices, via anonymous prompts, without putting the company at risk?

2. Sanctuarize time for practice

One hour a week, half a day per month — find the right rhythm and protect it. Practice should happen during working hours, not just on personal time.

3. Apply an “AI-first” reflex when talking to your team

If your team..

  • Asks for your opinion? Tell them to ask AI first.
  • Needs help structuring a project? Tell them to start with AI.
  • Wants to validate a business case? Compare together with AI inputs.

Show them it will not replace their thinking. It may not even "augment it", as we often read: but if will, for sure, help them be more confident in their own decisions.

4. Elevate your team through a concrete AI initiative

Even large-scale deployments should be iterative. Move beyond “one-off use cases” and articulate a clear ambition to leadership — while delivering step by step. That’s how you build both credibility and momentum.

5. Embrace the constraints of your organizations

They’re not always a bad thing. align with your Tech & Ops teams, and work with existing processes — not against them. Change happens faster when you bring others along.

Experimenting these recommendations is the first step towards a true mindset change in the way AI potential is understood by your team. It is also a prerequisite to delivering value at scale with AI, with motivated teams giving their best to grasp the new opportunities ahead.

And you - what have you put in place to help your teams adopt AI and learn through practice?

Crédit photo : AXA