AgentFarms v2 — 12 AI worker roles, approval gates & Azure isolation.See what's new
AgentFarms/Docs
Get API key
Documentation

AgentFarms Docs

Everything you need to deploy AI workers, connect real systems, and understand the governance model behind controlled AI execution.

New to AgentFarms?

Start with the Quickstart guide — you can have a worker live and processing real tasks in under 10 minutes. No credit card required.

Explore the docs

Quickstart

Deploy your first AI worker in under 10 minutes.

Get started →

Core Concepts

Understand the task lifecycle, execution model, and approval controls.

Read concepts →

REST API Reference

Programmatically manage workers, tasks, and approvals.

View API →

Connectors

Connect GitHub, Jira, Slack, Salesforce, and 14 more tools.

Browse connectors →

Approval Gates

How risk classification and human-in-the-loop review works.

Learn approvals →

SDK

TypeScript SDK for building on top of AgentFarms.

View SDK →

What is AgentFarms?

AgentFarms is a governed AI worker platform. It lets you deploy role-based AI workers that execute real tasks inside your tools — GitHub, Jira, Slack, Salesforce, and more — while routing important decisions through human approval gates. Every action is logged in an auditable evidence trail.

AgentFarms is not a chatbot, copilot, or code suggestion tool. Workers execute complete multi-step workflows autonomously: they open PRs, update CRM records, triage support tickets, write documentation, and coordinate across tools — with a governance layer that stops risky actions before they reach production.

How AgentFarms fits into your stack

AgentFarms connects to the tools your team already uses. Workers receive tasks from the dashboard, API, or trigger integrations (webhooks, email, Slack). They execute inside a scoped, isolated environment and surface outputs back through the same channels.

# Request flow
Dashboard / API / Webhook
↓
AgentFarms API Gateway (auth, billing, approvals)
↓
Agent Runtime (LLM dispatch, action tiers)
↓
Connectors (GitHub, Jira, Slack, Salesforce…)

Key concepts

WorkerA deployed AI agent with a specific role (e.g. Backend Developer), scoped tool access, and a configured approval policy.
TaskA unit of work assigned to a worker. Tasks move through Receive → Plan → Execute → Review → Iterate.
Approval gateA checkpoint that pauses execution and routes a decision to a human reviewer before a high-risk action executes.
Evidence trailAn append-only, searchable log of every action, approval decision, and output produced during task execution.
ConnectorAn OAuth-authenticated integration with an external tool (GitHub, Jira, Slack, etc.) that gives workers scoped access to actions.
Risk levelA classification (low / medium / high) assigned to each action that determines whether it auto-executes or requires approval.

Plans & limits

PlanWorkersTasks / monthPrice
Starter+22,000From $299/mo
Pro+510,000From $599/mo
EnterpriseUnlimitedUnlimitedCustom

All plans include a 14-day free trial with no credit card required. See full pricing details.

Next →Quickstart