KurnellKurnellAgents divide the work by role; people remain accountable for the judgement.
The first two matter most.
Not one general-purpose agent, but agents that each carry a role and the knowledge it needs.
Agents carry the execution; what to build, why, and when it is good enough stays with people.
How we work and how we verify goes on the site and into our writing — not just the results.
Repetitive work goes to automation so human time is spent only where judgement is needed.
Nothing grand — small attempts, repeated.
We look at problems worth productising, user situations, and gaps in the market.
Research turns into hypotheses and priorities.
A small prototype or a piece of writing shows whether the hypothesis holds.
What we learn is written up, so the next cycle starts from there.