Teaching an AI coach who it's talking to. Help the product understand the person.

01 · The problem

The feedback was useful. The product didn't know enough about the person.

Ren's AI could give good feedback, but it didn't always know enough about the person, their role, or the history of the relationship. Customer conversations and my own use of the product pointed to the same gap: missing context.

I worked with the founder and a small team over 90 days to decide what the product needed to understand, and to carry that into onboarding, profiles, manager views and shared context across web and Slack.

02 · What changed

Start with the person. Carry the context forward.

1. Bring the core idea forwardUse Ren's accountability model as the test for new features, so promising ideas don't pull the product in different directions.
2. Ask better questions at the startShip an onboarding profile that captures role, reporting lines, responsibilities and preferences.
3. Carry context across web and SlackWork with Ren's AI engineer so role, preferences and relationship history shape the experience on both surfaces.

03 · Deciding what deserves to exist

A thinking tool for a team moving quickly.

The team could explore and ship ideas fast. The harder work was deciding which ideas strengthened the coaching experience and which only made the product busier. So I built Anchor, a tool that asks whether a feature idea strengthens the core experience before momentum makes the decision for the team.