Profiles

Our agent team now works with profiles. When a founder or (human) team member joins, they pick how they work, engineer or product owner, and the way agents interact with them changes to match.

The same agents and the same goals, but the conversation, the gates, and the surface you work on reshape themselves around the person on the other end. One tool, two tones of voice.

How do the agents work with a product owner?

They abstract away the technical conversation. The workspace leans into design, strategy, and the outcome you’re trying to reach. Conversations focus on the experience: what it should feel like, who it’s for, what trade-offs are worth making. The agents return with production-ready results rather than implementation detail.

You don’t see code unless you ask for it. You see the feature, the flow, and the decisions that shaped it.

How do the agents work with an engineer?

As a technical peer. Every change gets code reviewed. Every architectural decision requires explicit approval before the agents commit to it. The conversation is dense, specific, and fluent in the stack: references to files, functions, and trade-offs sit where you expect them to.

Nothing ships past you that you didn’t sign off on. The agents are collaborators, not an autopilot.

Are profiles a permission system?

No. They’re a posture. You can switch yours any time in settings, and mixed teams work the way you’d hope: the same post, one engineer reviewing a diff, one product owner weighing in on the experience, both talking to the same agents in the voice that fits them.