For AI builders
Agents that know your full context.
Your pod is structured data infrastructure for both you and the agents you use. Any client. Full history; changes are reviewable and revertible.
Connect the agent you trust.
Synap is infrastructure for your agents, not a walled garden. Each client has a setup guide and a status we keep current.
Never locked to one provider.
Hub Protocol is an open REST contract, and every pod runs an MCP server. Any agent or intelligence service that speaks them can connect to your pod.
Switch models without losing context. Run different models for different tasks. Self-host a model on your own hardware and connect it to your Synap pod.
Your data doesn't move. The intelligence is pluggable.
Agents that actually know you.
Most assistants have no memory of your work. Every conversation starts from zero.
Your Synap pod stores everything: notes, tasks, projects, people, articles, files — with relationships between all of them. The agent sees your full entity graph, not just a chat history.
Ask “what did I learn about distributed systems last month?” and get a real answer.
Agents propose. You decide.
By default, agents in Synap don't modify your data directly. A consequential write becomes a proposal — a structured diff with reasoning that explains what change is suggested and why.
You see it. You approve or reject. Only then does anything change. For an agent you trust, you can auto-approve specific kinds of change — a rule you set and can take back.
Full history; changes are reviewable and revertible. This is what “human in the loop” actually means.
Specialized agents for complex tasks.
Complex requests are decomposed by an orchestrator that dispatches to specialists: Knowledge Search, Code, Writing, Action, Insight Discovery, View Builder, Workspace Builder.
Each agent has its own tool set and constraints. Knowledge Search traverses your entity graph. Writing works from your document history.
You always see what's happening and why.