Platform
Run conversations, keep users and credentials apart, track usage, and handle failures in production.
Everything around the agent that you need in production: conversations that persist, per-user reporting, safe credentials, usage limits, billing your own users, and chat apps. Start with sessions if you are building a chat, and read the rest before you go live.
Explore the platform
Sessions and turns
Keep a conversation going across requests and read its history.
Tenancy and attribution
Label activity with your own user IDs to filter and report per user.
Security and credentials
Keep your API key on your server and rotate it safely.
Usage and quotas
See what every turn used and set monthly safety ceilings.
Bill your users for model tokens
Charge your customers for tokens through your own Polar or Dodo account.
Limits and reliability
Retry safely, handle cancellation, and debug failures by symptom.
Slack and Telegram
Put your agent in Slack or Telegram with your own bot.
How a request flows
- Your backend signs in the user, checks what they may access, and calls Blazing Agents with your API key.
- You can pass a
userIdto label the work for reporting. It does not grant access. - The agent runs a turn, inside a session if you want history, and its usage is recorded.
- Quotas and service limits decide whether the turn starts, and the error code tells you whether to retry.
Next
- Agents for what an agent can use and produce.
- Automation to run agents in the background or on a schedule.