AI

Bridge your ecosystem to AI. Connect application states and human-curated data directly to agents through the HTTP API (and MCP where you expose BigState tools)—so models and workflows get live data for analytics, training signals, and automation.

Apps + humans publish state → BigState holds the live source → agents read / subscribe → act

Why BigState for AI#

Live data source

Agents always see the latest versioned state—not a stale export file.

How agents consume state#

  1. Authenticate as a dedicated principal with an API key or token.
  2. Read latest state (or a specific version) for the objects the agent needs.
  3. Subscribe via WebSocket delivery when the agent should react as values change.
  4. Resolve valueRef through the Value API when the state points at files or images.
  5. Act in your agent runtime (tool call, workflow, training job)—BigState remains the shared source of truth.

If you wire BigState into an MCP server or other tool layer, expose the same read/list/subscribe operations your HTTP client already uses so agents call one consistent model.

Where to go next#

Related#

  • Applications — services that produce the states agents use
  • Humans — curated public reports and files in the same model
  • Tutorial — end-to-end apps you can mirror in agent tooling
HumansUse cases

© 2024 BigState