We have the first real chance in decades to redesign cognitive tools from first principles. LLMs absorb complexity and reduce cognitive load, the same qualities that make them good for coding can transform how we read, write, research, and think.

But this requires intention.

Left to default incentives, AI tools drift toward engagement. The alternative is opinionatedly designed tools that protect focus, respect agency, treat sustained thinking as the goal.

We're building tools that optimise for cognition. Not time-on-task, not engagement, not output volume.

We think designing for how people actually think produces better, more sustainable work.

In practice, this means reducing context-switching, building AI that assists judgement rather than replaces it, and treating boundaries and reflection as core features rather than afterthoughts.

Success gets measured by what you accomplish, not how long you stay.