Three Takeaways from the INSEAD AI Forum Americas 2026
Last week, I had the opportunity to attend the INSEAD AI Forum Americas 2026 at the INSEAD San Francisco Hub for Business Innovation.
One thing I particularly enjoyed was the format of the event. Beyond the sessions, there was a strong focus on networking and exchanging perspectives with people from different industries and backgrounds who are facing many of the same questions around AI. I left with great conversations, new connections and plenty of ideas to think about.
A technical topic that caught my attention: System Prompts
We talk a lot about models and AI agents, but much less about the system prompts that define how those systems are expected to behave.
One question from the discussion particularly stayed with me:
“We can define how an AI should behave. But how do we verify that it actually behaves that way?”
As agents become more autonomous, this will become an increasingly important part of AI governance.
Beyond this specific topic, three broader ideas stayed with me.
1. Start with the problem, not with AI
AI is a solution, not the starting point.
Instead of asking “Where can we use AI?”, organizations should first identify the business problem, need or bottleneck they want to solve. Only then should we decide whether AI is the right tool and how it can create real value.
With the current excitement around agents and new models, it is surprisingly easy to do the opposite: start with the technology and then look for a problem.
2. The next enterprise AI challenge is coordination, not more agents
We are quickly moving toward organizations with specialized AI agents across sales, finance, operations, legal and other functions. But having more agents does not automatically create a better organization.
The real challenge is what happens between them: how they share context, coordinate decisions and interact with people and enterprise systems.
One particularly interesting point from the sessions was that deploying specialized agents without redesigning the overall workflow could actually reinforce existing organizational silos rather than remove them.
For me, this is where the enterprise AI conversation becomes particularly interesting:
“Not more agents, but better orchestration.”
3. Redesign the organization to turn People + AI into business value
This was my biggest takeaway.
The opportunity is not simply to automate what people do today or give everyone an AI assistant. It is to rethink how work gets done when people and AI work together.
That means reconsidering workflows, responsibilities and decision making, while understanding what should be automated and where human context and judgment remain essential.
From a CIO perspective, this changes the question from “Which AI should we deploy?” to “How should we redesign the way our organization works with AI?”
As access to powerful AI becomes increasingly widespread, simply having the technology will not be enough to differentiate an organization.
I left the INSEAD AI Forum with new ideas, great conversations and one main reflection:
“The next phase of enterprise AI is not just about adopting better models or adding more agents. It is about redesigning how People + AI work together to create real business value.”
