Voca AI

Every conversation with Voca is mined for goals, decisions, blockers, and progress. Those get sorted into personal, project, and workspace layers of memory. The point is that you don't have to tag anything or write a summary yourself, since it's extracted as you talk. Over time the knowledge base grows on its own rather than needing manual maintenance.

Voca separates knowledge into three tiers: what's true for you, what's true for a project, and what's true across the whole workspace. That structure matters for teams. A decision that affects one project shouldn't clutter everyone's view. It's the mechanism that lets context persist across sessions instead of resetting each time.

The assistant connects to GitHub, Linear, and Slack through MCP, so it reads your work where it already lives rather than asking you to re-enter it. Once connected, Voca can observe commits, tickets, and threads and fold them into its understanding of a project. The trade-off is clear. Fewer connected tools means less useful context.

You can define custom Skills, meaning the specific checks or routines you want, and Heartbeat runs them on a schedule. This turns Voca from a passive observer into an automated teammate. The recurring work just happens. It's the feature that most resembles a traditional automation tool, but tied to the context Voca has already gathered.

When something important happens, Voca decides who needs to know and notifies them, instead of waiting for a human to forward the news. That's the "invisible PM" idea in practice. The routing happens in the background, so no one has to play messenger. It works best when team roles and tools are already well defined.

Voca combines activity across all connected projects and surfaces a short recommendation of what to focus on today. It's meant to cut the mental work of scanning every tool just to figure out priorities. That scan gets old fast. The suggestions are only as good as the data flowing in, so teams that connect fewer tools will see thinner results.