The 7 Human Skills That Matter More in an AI World
I’ve been asked a version of the same question for the past three years: which jobs in our innovation team are at risk from AI? My answer has stopped changing: it’s not about jobs. It’s about which capabilities your team is building — and which ones are quietly atrophying because AI is handling them.
I work with FMCG teams who are genuinely good at integrating AI into their workflows. They move faster. Their research preparation is sharper. Their concept articulation is more polished. And in several cases, they’re becoming less good at the things that AI can’t do.
These are the seven capabilities I’ve seen make the real difference in innovation teams over 20 years — and the ones I watch most carefully when I’m coaching.
Reading a room, not just a dataset
The most important insight I’ve encountered in 20 years of innovation work has never come from a survey or a report. It came from watching someone’s face change when they held a product, from a hesitation in a consumer interview that contradicted everything the person had just said.
AI can analyse sentiment in text. It cannot observe the gap between what someone says and what they do. That gap is where the most useful insights live — and it requires a human to notice it.
Asking the question that stops the room
Innovation sessions go wrong when they follow the path of least resistance — when teams build on what they already believe rather than questioning it. The most valuable thing a person can do in a room full of smart, aligned colleagues is ask the question nobody wants to answer.
“Are we solving the right problem?” “Who is this actually for?” “What would have to be true for this to fail?”
These questions cannot be generated by AI — because AI doesn’t know what the team is avoiding.
Tolerating ambiguity long enough to find the real insight
There is a strong organisational pressure to move quickly from observation to conclusion. AI accelerates this — it can summarise a dataset in seconds and generate a recommendation.
The problem is that the most valuable insights in innovation rarely arrive quickly. They emerge from sitting with contradictory data, from resisting the first obvious interpretation, from letting two apparently unrelated observations connect over time.
I’ve seen teams shortcut this process with AI summaries and arrive at technically accurate but strategically shallow conclusions. Tolerance for ambiguity is not inefficiency. It’s part of the work.
Building alignment across functions — not just communicating downward
Innovation is 10% idea and 90% alignment. I’ve watched genuinely strong concepts die in FMCG because R&D, Marketing, and Operations never reached a shared understanding of what they were trying to do.
Getting a cross-functional group to genuinely agree — not just nod in a meeting — requires mapping each function’s real concerns, finding the formulation that makes the concept coherent from every angle, and sometimes having a direct conversation about why someone is blocking.
AI can help draft the alignment document. It cannot have that conversation.
Knowing when to stop
I’ve written more about this in the article on Pivot, Persevere, and Stop. But the underlying capability worth naming here is the willingness to close something you’ve worked hard on when the evidence says it’s not working.
This is genuinely difficult in organisations where people’s careers are tied to the projects they’ve championed. It requires separating the quality of the idea from the quality of the judgment that stops it. And it requires a leader who makes it safe to stop — which is a deeply human act.
Translating between technical and commercial language
In almost every FMCG innovation team I’ve worked with, there is a translation problem. R&D speaks one language, Marketing speaks another, and the gap between them is where time and money get lost.
This translation is not just linguistic. It requires genuinely understanding both sides — what a flavour profile means for a production process, what a consumer trend means for a formulation constraint. I spent the early part of my career on the technical side, which means I can hold both conversations at once.
That cross-functional literacy is rare and it’s not teachable in a workshop. It comes from having been in both rooms.
Coaching people who are better than you at their specialty
The best innovation coaches I’ve encountered are not the smartest people in the room on any given topic. They are the people who can help the smartest person in the room do their best thinking.
This means asking questions rather than providing answers. It means knowing when to push and when to get out of the way. It means being genuinely curious about a colleague’s expertise rather than performing curiosity.
AI can generate questions. It cannot model the intellectual humility that makes those questions land.
A note on what this isn't
This article is not an argument against AI in innovation teams. I use it daily, and I’ve written about exactly how in a separate piece on AI in FMCG workshops. The point is different: the teams that will work best with AI are those that have invested in the capabilities AI cannot replace.
The most useful innovation professionals in the next decade will not be the ones who delegate most to AI. They will be the ones who know what to delegate — and what to hold.
“The skills that matter most in an AI world are not the technical ones. They are the ones you could only have developed by being wrong in a room full of people — and learning from it.“
Amandine Devergies
If this resonates
I’ve applied these ideas across FMCG teams in Europe, Asia, and North America — with different categories, different constraints, and different levels of AI maturity. If you’d like to see how this plays out in specific project contexts, have a look at my projects section. Or connect with me on LinkedIn.
4 Responses