Insights

What we've learned.

Reports, working notes, and commentary on AI implementation.

Most firms think knowledge capture means recording meetings. It doesn't. Here is what it actually looks like.

Large context windows sound like a silver bullet. They are not. How you structure what goes in matters more than how much fits.

A year ago, everything seemed to hinge on clever prompts. It does not. Stable AI systems need engineered context: data, memory, tools, rules.

The shift from AI that helps you type faster to AI that completes entire tasks changes what it means to be good at your job.