AI Deck

Flowise — An Open Source Platform for Building AI Agents by Drag and Drop (Service Ends August 2026)

Flowise is an open source low-code platform that lets you assemble AI agents, RAG (retrieval-augmented generation) pipelines, and chat assistants by dragging and dropping nodes onto a canvas in the browser. Since its release in April 2023, it has drawn more than 50,000 GitHub stars on the strength of over 300 nodes built on LangChain and LlamaIndex integrations, plus instant deployment through APIs and embeddable widgets. In August 2025, Workday acquired the company and folded the technology into its own AI agent platform.

At the same time, the project announced its sunset on July 29, 2026. The code was frozen that same day, the GitHub repository was archived on August 13, and end of life (EOL) arrives on August 31, 2026. The source code remains publicly available under the Apache 2.0 license, so you can fork it and keep using it on your own, but official updates and support are ending. If you are considering adopting it now, that fact needs to be part of the decision.

Key Features

  • Visual flow design: Two modes are available ─ “Chatflow” for building single-agent chat assistants, and “Agentflow” for orchestrating multiple agents. Branching and loops can also be assembled on the canvas
  • 300+ nodes and LLM integrations: By incorporating LangChain and LlamaIndex components, you can connect more than 100 LLMs and embedding models, vector databases, and external tools as nodes
  • RAG pipeline construction: Document loading, splitting, embedding, and retrieval can be designed as a single flow, letting you build an assistant that answers from internal documents without code
  • Human In The Loop (HITL): A mechanism for a person to review and approve an agent’s decisions mid-run, making runaway automation easier to prevent
  • Observability and debugging: Execution traces can be followed step by step, with metrics export to Prometheus and OpenTelemetry
  • Multiple deployment paths: REST API, embeddable chat widget, and TypeScript/Python SDKs are provided. Both self-hosting (Docker, npm, various clouds) and managed cloud are supported

Pricing

PlanMonthly PriceMain LimitsKey Features
Free$02 flows and assistants, 100 predictions/month, 5MB storageEvaluations, custom branding, community support
Starter$35Unlimited flows, 10,000 predictions/month, 1GB storageEverything in Free, community support
Pro$6550,000 predictions/month, 10GB storage, unlimited workspacesAdmin roles, 5+ users ($15 per additional user), priority support

The self-hosted edition costs nothing for the software itself; you pay only for servers and LLM API usage.

Pricing is as of August 2026. With the service ending on August 31, 2026, availability and terms for the cloud edition may change. Please check the official site and the sunset notice for the latest information.

Pros & Cons

Pros

  • Complex setups such as multi-agent systems and RAG can be designed visually without writing code
  • Apache 2.0 open source, so self-hosting keeps your data inside your own environment
  • With 300+ nodes, you can swap major LLMs, vector databases, and external tools in and out to compare them
  • APIs, SDKs, and an embeddable widget are all available, carrying you from prototype to integration in your own product
  • Features aimed at production use, such as execution traces and HITL, are included from the start

⚠️ Cons

  • Official service ends on August 31, 2026, after which there will be no updates, security fixes, or support
  • The GitHub repository was archived on August 13, 2026, so issues and pull requests are no longer accepted, and the npm packages and Docker images are marked deprecated
  • Continuing to use it requires forking and maintaining it yourself, which effectively assumes you have development resources to spare
  • The sheer number of node types makes the first flow hard to design without a working knowledge of LLMs and vector databases
  • When self-hosting, servers, databases, and scaling are your own responsibility

Comparison with Similar Services

CriteriaFlowiseDifyLangflown8n
Primary useAI agent and RAG buildingLLM app development and operationsRapid AI workflow prototypingGeneral workflow automation plus AI
LicenseApache 2.0 (archived)Open source (with some restrictions)MITSustainable Use License
Self-hostingYes (maintenance on you)YesYesYes
AvailabilityEnds late August 2026OngoingOngoingOngoing
Best suited toExisting users and fork-maintaining teamsTeams running internal LLM apps in productionDevelopers who want fast prototypesTeams automating across business systems

If you are choosing a tool of this kind from scratch, Dify, Langflow, and n8n are the realistic candidates because they are still under active development. All of them support visual flow design and self-hosting, so the design thinking behind a Flowise setup carries over almost directly.

Who Is It For

  • People already running production flows on Flowise who need to decide how to continue in the near term
  • Development teams able to fork the Apache 2.0 code and keep maintaining it in house
  • Anyone who wants to learn agent architecture and RAG pipeline design from a working open source codebase
  • People interested in AI agent building on the Workday platform who want to understand where it came from

If you are looking for a no-code AI agent platform for a new project, it is better to start with options that are still supported.

Summary

Flowise grew into a leading open source platform for building AI agents and RAG by drag and drop, gathering more than 50,000 stars along the way. After the Workday acquisition, its official service ends on August 31, 2026. The code lives on under Apache 2.0, so it retains value as learning material and as a base for forks, but it is not suited to new adoption where support is a requirement. Existing users should weigh running a fork against migrating to another tool, while newcomers should look at alternatives that are still maintained.

← Blog