Activepieces
Open-source, AI-first automation platform with no-code AI agents and MCP
Last reviewed 2026-06-20
Activepieces is an open-source workflow automation platform, positioned as a self-hostable alternative to Zapier and Make, with AI agents and Model Context Protocol (MCP) built in. A visual no-code flow builder chains triggers, integrations, conditions, loops, and code into automations across 700+ apps (called 'pieces'), and every piece can also be exposed as an MCP tool for AI assistants like Claude, Cursor, and Windsurf. The AI agent builder lets non-technical teams configure agents that pursue a goal by calling connected tools, with built-in human-in-the-loop approval ("agents ask when they need you") and full run logging. The Community Edition is MIT-licensed and free to self-host; a managed cloud and an embed product target teams and SaaS vendors who want automation inside their own apps.
What it can do
Build no-code AI agents that call tools
SupervisedThe AI agent builder lets users configure an agent that pursues a goal across connected apps, choosing and calling tools to complete the task, with built-in approval steps for actions that need human judgment.
sourceOrchestrate multi-step automations visually
AssistantA visual flow builder chains triggers, integrations, conditions, loops, auto-retries, HTTP requests, and code (with NPM packages) into workflows that run on schedule, webhook, or event.
sourceExpose integrations as MCP tools
SupervisedEvery Activepieces integration can be served as an MCP tool, letting AI assistants such as Claude, Cursor, and Windsurf read, act, and trigger flows; the project describes itself as one of the largest open-source MCP server collections (~400 servers).
sourceHuman-in-the-loop approval
CopilotAgents can pause and ask for human approval on consequential actions (the docs give a refund outside the standard window as an example), and every run is logged with full input/output per step.
source
Strengths
- +Genuinely open source: the Community Edition is MIT-licensed and free to self-host with no task limits
- +AI-first design: no-code AI agents plus native MCP, so every integration doubles as a tool for AI assistants
- +Large and community-driven piece library (700+ apps, ~60% of pieces contributed by the community)
Limitations
- −Enterprise security controls (SSO, RBAC, audit logs, Git sync) are gated behind the custom Ultimate/enterprise tier
- −Self-hosting shifts backups, security updates, scaling, and infra cost onto you
- −AI agents are supervised, not fully hands-off: consequential actions can require human approval
Overview
Activepieces is an open-source, AI-first automation platform, positioned as a self-hostable alternative to Zapier and Make. It combines a visual no-code flow builder, a no-code AI agent builder, and native Model Context Protocol (MCP) support. The Community Edition is MIT-licensed; the project reports 22.9k+ GitHub stars as of June 2026. It was co-founded in 2022 by Ashraf Samhouri and Mohammad AbuAboud and is a Y Combinator company.
What it does
The flow builder chains triggers, integrations, conditions, loops, auto-retries, HTTP requests, and code (with NPM packages) into automations across 700+ apps ("pieces"). The AI agent builder lets non-technical users configure an agent that pursues a goal by choosing and calling connected tools, with approval steps for actions that need human judgment and full per-step run logging. Activepieces also serves its integrations as MCP tools, so AI assistants like Claude, Cursor, and Windsurf can read, act, and trigger flows; the project describes itself as one of the largest open-source MCP server collections (~400 servers).
Integrations & setup
700+ pieces including OpenAI, Anthropic Claude, Slack, Gmail, Notion, HubSpot, Google Sheets, and an HTTP node for any API. Roughly 60% of pieces are community-contributed and published on npm. It is model-agnostic across GPT and Claude. Deployment options are managed cloud, self-hosted Community Edition, on-prem, and an embed SDK for putting the builder inside your own product.
Pricing
The self-hosted Community Edition is free with no task limits. Cloud is free for up to 10 active flows with unlimited runs, then $5 per active flow per month; the Ultimate / enterprise tier (SSO, RBAC, audit logs, Git sync, private pieces) is custom-quoted, and the embed product is quoted separately.
Best for / not for
Best for teams that want open-source, self-hostable automation with first-class AI agents and MCP, and for SaaS vendors embedding automation in their own apps. Less suited to those who want a fully managed, no-ops product with zero infrastructure responsibility, or who need turnkey enterprise security controls without an enterprise contract. As with most agent builders, treat the agents as supervised: consequential actions can still require human approval.
Alternatives
n8n is the closest peer (workflow automation with AI agent nodes and MCP, but fair-code rather than MIT-licensed). Make and Zapier are the larger commercial automation incumbents. Pipedream and Gumloop overlap on developer-leaning and AI-native automation respectively.
What people are saying
We aggregate real LinkedIn discussion into sentiment for the agents people search most. Activepieces isn't tracked yet, want it added? Request tracking.
FAQ
Is Activepieces open source?+
Yes. The Community Edition is MIT-licensed and free to self-host, with enterprise features released under a separate commercial license. This makes it more openly licensed than n8n, which is fair-code / source-available rather than OSI-open.
Can Activepieces run AI agents?+
Yes. Its no-code AI agent builder lets an agent pursue a goal by calling connected tools across 700+ apps, with built-in human approval steps and full run logging. Separately, every integration can be exposed as an MCP tool for assistants like Claude, Cursor, and Windsurf.
How much does Activepieces cost?+
Self-hosting the Community Edition is free. Cloud is free for up to 10 active flows with unlimited runs, then $5 per active flow per month; enterprise (Ultimate) pricing is custom, and the embed product is quoted separately.
Sources
- Activepieces GitHub repository (MIT license, pieces, MCP, stars) · accessed 2026-06-20
- Activepieces AI Agent Builder product page · accessed 2026-06-20
- Activepieces MCP Server documentation · accessed 2026-06-20
- Activepieces pricing · accessed 2026-06-20
Last reviewed 2026-06-20