06 · Product
AI Agents
The model is machine-readable. AI agents act on it.
An agent is only as good as what it can see and what it can do. In synapse™ it sees the whole system model, the requirements, the interfaces and the workflows, and it can act on all of it: answer a question, make the edit, run the calculation.
- It does the work, not just the answerAsk whether a requirement still holds after a change, or ask for the change itself: a new interface, a reparented requirement, a workflow built and run.
- Under your permissions, behind the same reviewThe agent reaches only what your account can. Anything destructive waits for your approval, and its edits become baseline through the same merge request a person's would.
One instruction: the agent raises battery_capacity to 180 Ah, re-runs the workflows that depend on it, re-checks every requirement against the new numbers and bundles the result as a merge request for a person to approve.
Integrations
Bring the agent you already use
Over MCP, an external client gets the same tools as the built-in agent, on the live project, under the same permissions and the same review.
- Claude
- ChatGPT
- Gemini
- Cursor