AgentsAn agent is a
An agent is a
configuration with history.
Give it instructions, a model, tools and knowledge. Every change archives a version, so what answered yesterday is still there to read and roll back to.
What it does
What agents do
01
Versioned by default
Every write that changes the configuration archives a version, and a release can serve two versions side by side for a staged rollout.
02
Tools, guarded
Bind tools the agent may call, choose which are active, and put guardrails and approvals on the calls that need a person.
03
Grounded in your documents
With knowledge set, retrieval runs before every generation, so answers come from what you ingested rather than what the model remembers.
04
Any model, with failover
Pin a provider, or name a model route that resolves a model and fails over when one is down.
05
Structured when code reads it
Attach a JSON Schema and every non-streaming generation returns a validated object instead of prose.
06
Read back, step by step
Every generation leaves a transcript, kept for your plan's retention window, of what the agent did, which tools it called and what it cost.
Tutorial
Your first agent generation
By the end you will have an agent that answers — created, run once, and read back step by step. Every step is one call, shown for the CLI, the SDK and curl.
Start the tutorialMore features
More features
- KnowledgeGround answers in your own documents, with the source to cite.
- ChannelsPut an agent on WhatsApp or Discord, where people already are.
- AutomationRun agents on a schedule, an event or by hand, with a record of each firing.
- ControlGuardrails on tool calls, human approvals and an audit log.
- EvaluationsScore a change against your own cases before it ships.
- ModelsManaged models with no vendor account, or your own credential.
Build the agent.
Keep the receipts.
Sign in, connect your assistant or open the quickstart, and run your first agent.
Free plan · 2,000 runs a month · no card
