The framework
AI is not a
productivity tool.
It is a multiplier.
Multipliers amplify what they find. In a product team with strong foundations, AI sharpens judgement, surfaces what people were afraid to say, and accelerates decisions worth making.
Without those foundations, it accelerates the wrong things faster. With higher confidence. And less visibility of the problem.
Care capital is not a culture initiative. It is a product quality issue. Five measurable conditions determine whether a team can exercise genuine collective judgement, use AI tools well, and ship work that matters.
Teams that build it ship better work. Teams that don't will be faster at producing the wrong thing.
The argument
The teams that handle AI well are not the ones who figured it out alone.
They are the ones who already had the conditions in place: people who could challenge each other honestly, shared a mission they believed in, and built on each other's thinking rather than just coordinating outputs.
Care capital is the name for those conditions. It exists in every team. The question is whether you have enough of it to do what you are being asked to do.
The five conditions
What Care Capital measures.
Five interdependent conditions. Each affects the others. The order matters: safety enables trust, trust enables aligned incentives, and so on until you reach the only output that counts.
01 Institutional Safety Whether people can raise real concerns without career cost ▼
In most product organisations, the stated culture and the lived experience are not the same thing. People learn quickly what can be said and what must be managed.
In AI-assisted work this matters more than ever: AI surfaces what people have been too careful to say out loud, and it takes genuine safety to act on it rather than dismiss it.
- People challenge product decisions without it becoming political
- Mistakes are examined rather than managed away
- Leadership behaviour matches what leadership says
- Teams ship what is safe rather than what is right
- AI outputs get rubber-stamped because no one wants to question the model
- Critical feedback loops break down quietly
02 Team Trust & Empathy Whether product, design and engineering understand each other ▼
Cross-functional product teams are asked to collaborate under pressure, across different professional languages, with different definitions of done.
Without trust and empathy, AI tools become individualised. Decisions get made without the full picture. The team fragments into capable people working in parallel rather than together.
- Designers and engineers understand what motivates each other
- Disagreement is navigated rather than avoided or escalated
- People know where their teammates find things difficult
- AI adoption happens individually, creating invisible capability gaps
- Collaboration becomes coordination: lots of communication, little shared thinking
- The team's combined judgement drifts toward the lowest common denominator
03 Incentives & Mission Whether the team is pointed at outcomes it can actually influence ▼
Product teams are frequently held accountable for lagging indicators they cannot directly control. When AI is introduced, this compounds: the team can move faster, but faster in the wrong direction is worse.
A team with well-structured incentives uses AI to go further toward outcomes that matter. A team with misaligned incentives uses AI to hit numbers that do not.
- The team's success metrics are things it can actually move
- There is a shared definition of what a good outcome looks like
- Collective achievement is recognised, not just individual performance
- AI gets used to optimise metrics rather than to improve outcomes
- Short-term velocity crowds out long-term quality
- People build what they are measured on rather than what users need
04 Curiosity & Growth Whether the team has space to challenge the brief and develop ▼
Teams under delivery pressure lose curiosity first. By the time they notice, the habit is gone. In an AI-augmented workflow, the teams that thrive are the ones that stay curious about the problem, the tools, and whether the assumptions are right.
Curiosity is what turns AI from an automation layer into an amplifier of insight.
- People are encouraged to ask questions, even when they slow things down
- The team challenges its own assumptions, not just the brief
- There is real investment in developing people, not just upskilling them for current tasks
- AI becomes a shortcut to the obvious answer rather than a tool for better questions
- Everyone's output starts converging toward the AI's average
- The team stops learning because it is too busy delivering
05 Collective Intelligence Whether collaboration produces better decisions than any individual alone ▼
Collective intelligence is the output condition: the thing all four previous conditions make possible. When they are healthy, the team produces decisions and work that none of its members could have reached alone.
AI can either accelerate collective intelligence or substitute for it. The difference depends entirely on what the team brings to the collaboration.
- Decisions made together are better than decisions made individually
- People regularly change their minds because of something a colleague said
- Working with this team makes each person's own thinking sharper
- AI produces confident, coherent outputs that feel like collective thinking but are not
- The team loses the ability to disagree productively
- Group judgement degrades below that of its strongest individual members
Free diagnostic
Where does your team sit?
Twenty-five questions across the five conditions. Eight minutes. You get a radar chart of your scores, benchmarked against other product teams, with specific recommendations for where to focus.
Complete it individually, then share with your team. The variance between responses is often the most useful data of all.