← Back to stack

Builder infrastructure

Flowise

Build AI Agents, Visually

Visit official site →

Flowise is an open-source, visual agentic-systems development platform: a drag-and-drop builder for chat assistants, single-agent chatflows, and multi-agent "Agentflow" orchestration, deployable self-hosted or via Flowise Cloud.

Vendor's own claim

What Flowise is

Vendor's own claim

Flowise is a visual, node-based development platform for building LLM-powered chat assistants and agentic workflows without writing the underlying orchestration code by hand. It layers three builders of increasing power (Assistant, Chatflow, Agentflow) over a shared engine, exposes the result via API/SDK/CLI, and documents both an MCP client and server integration. It can be installed locally via npm/npx, run via Docker, or used as a hosted Flowise Cloud product.

What it can do

Vendor's own claim
  • Three visual builders: Assistant (beginner-friendly agent), Chatflow (single-agent/RAG flows), Agentflow (multi-agent orchestration with branching/looping)
  • Human-in-the-loop review step within agent workflows
  • Execution tracing and observability, with integrations named for Arize, LangWatch, Langfuse, Lunary, Opik and Phoenix
  • MCP client/server nodes with tool listing and SSE/auth support
  • REST API, JS/Python SDK, CLI, and an embeddable chat widget
  • Self-hosted (including air-gapped) or Flowise Cloud deployment
  • Connects to 100+ data sources, tools, vector databases and memory backends
  • RBAC, SSO, encrypted credentials, secret managers, rate limiting and restricted domains for security controls

Who it is for

Our read

Developers and technical teams who want to prototype or run LLM agent workflows visually rather than hand-coding an orchestration layer, and who are comfortable self-hosting or using a small managed cloud instance.

When to choose something else

Our read

Self-hosting (the path the docs lead with) requires comfort with Node.js, Docker and ongoing maintenance; teams that want a fully managed, zero-ops product without engineering involvement are working against the product's default setup path.

Implementation considerations

Our read

Setup assessment: 1h

Quick Start is a two-command local install (`npm install -g flowise`, `npx flowise start`) after Node.js is present, or a Docker Compose setup with an `.env` file - both are fast for a first run, but the docs explicitly note self-hosting "requires more technical skill" for backups and ongoing updates, and building real Chatflow/Agentflow pipelines takes materially longer than the initial install.

Prerequisites

  • Node.js v18.15.0 or v20+ (for local/npm install)
  • Docker and Docker Compose (for the Docker path)

Verify before you adopt it

  • Details and timeline behind the 'sunsetting Flowise' banner (the linked 'Learn more' page was not in the confirmed set)
  • MCP auth scheme, base URL, and specific tool names
  • API authentication scheme
  • A machine-readable OpenAPI specification was not included in the reviewed source set.
  • Published pricing was not available in the reviewed source set; confirm current commercial terms with the vendor.