Activepieces

Activepieces is an open-source automation platform that combines visual flows, AI agents, chat, tables, and more than 760 app integrations in one workspace. It competes directly with Zapier Central, Zapier, and Make, but its MIT-licensed integration pieces, self-hosting path, and built-in MCP support give developers more control than a conventional closed SaaS connector catalog. Teams exploring agent workflows should also read the MCP builder guide.

The flow builder covers branching, retries, wait points, subflows, error handling, and human approvals. A user can keep common steps visual and add inline TypeScript when a workflow needs custom logic. The HTTP piece can call REST or GraphQL endpoints with authentication and pagination, reducing the need to wait for a dedicated connector. Connections are reusable across flows, agents, and teammates. Tables provide shared state that workflows and agents can read and write, while searchable documents can ground agent responses. Activepieces can consume external MCP servers and publish flows through its own MCP server, making an existing automation callable from compatible assistants.

Current pricing is unusually easy to model. Free costs $0 and provides 100 credits each day without a card. Plus is $16 per month when billed yearly and includes 10,000 credits monthly. Team is $166 per month billed yearly with 50,000 monthly credits and email support. Ultimate uses custom annual credits and enterprise terms. The vendor says one credit covers an entire ordinary flow run regardless of step count; AI use is different. A fast model step costs 2 credits, a smart model step 10, a frontier model step 20, and an agent action 1 credit. Plus and Team overages cost $0.007 per credit. Buyers can connect their own AI provider keys on Plus and above, which changes where model costs appear.

That credit model is a genuine differentiator from services that meter every workflow step. A 12-step deterministic flow can still consume one run credit, but an agent that loops through tools may use multiple action credits. Build a cost test with real branching and failure behavior rather than multiplying happy-path runs. Self-hosting can eliminate the hosted subscription for Community Edition, but it does not eliminate compute, database, upgrades, backups, secrets management, monitoring, or staff time.

A useful pilot is an approval-heavy process such as invoice intake: monitor an inbox, extract fields, match a purchase order, write a row to a table, request human approval, then update accounting software. Record completion rate, duplicate prevention, p95 duration, manual interventions, retry behavior, connector failures, and credits per successful outcome. Test credential rotation and least-privilege scopes before production.

Pros include transparent entry pricing, unlimited flows on every plan, a large open-source connector library, visual-to-code escape hatches, and strong MCP alignment. Cons include variable AI credit consumption, operational work when self-hosting, a smaller ecosystem than Zapier, and the need to inspect community-built pieces before granting sensitive permissions. Activepieces is a strong fit for technical operations teams that want low-code speed without surrendering deployment and extension options.