About
We're building the tool we wished we'd had.
blayd started inside a different company. We were a small team shipping AI features into an existing product, and we kept finding ourselves between four tabs (Jira, Notion, Slack, Drive) trying to give the model enough context to be useful. The model could write the code. It just couldn't see the work.
We tried wiring an agent against MCP servers, against vendor APIs, against Zapier-shaped integration layers. Every path led to the same dead end: someone, somewhere, was hand-rolling translations between four schemas, and the agent's knowledge of state was always one round-trip behind reality.
So we went the other direction. We built a workspace where the agent sees the same REST surface the human sees. One database. One auth context. One audit log. The agent's tool catalog is generated from the OpenAPI spec. There is no second backend to keep in sync, because there is no second backend.
That's blayd. It's a tool for the next ten years of work, not a wrapper around the last ten.
What we believe.
- 01
Software should be a place, not a pile of forms.
Tickets without docs, docs without channels, channels without files. Every modern tool has cut work into too many silos. blayd reads as one workspace because it is one.
- 02
The agent sees what the human sees. Nothing else.
No shadow APIs, no scraped state, no MCP middlemen. The same REST surface the desktop uses is the surface the agent uses. Same auth, same audit log.
- 03
Local-first. Your data is yours.
blayd runs on your machine, against a server you also run. Bring any model vendor. We never see your workspace contents.
- 04
Beta means beta.
We talk to every early user, we ship weekly, we tell you what is broken before you find it. The roadmap is shaped by what real teams hit, not what investors ask in pitches.
Where we are.
A handful of us, scattered across London, Berlin, and Lisbon. Quietly shipping. Talk to us at hello@blayd.app.