Flowise alternatives after the shutdown (2026)

What happened to Flowise

The code stays on GitHub under Apache-2.0, and the team suggests forking it. Its reason for stopping: coding agents now handle the complex AI work that visual low-code flows struggled with. We found no official date for Flowise Cloud shutting down, so export your flows now rather than later.

An archived Flowise keeps running, but it gets no security fixes. Keep a self-hosted instance off the public internet until you’ve moved.

Where to go, by what you built

1. Lazarus: our top pick

Best for business processes your flows were running, moved into one system your company owns

A builder where you describe your business and an AI designer designs your company’s own system: mail, phone, CRM, records and AI teammates that do the repetitive work inside it. It doesn’t import Flowise flows: we rebuild the process, not the JSON.

Price:

Designing it is free. Managed hosting from $150 a month; developer support from $4,500 a month.

Licence:

Commercial; the system built for you is yours

Self-host:

Managed hosting

Pick it when:

your Flowise flows were doing company work, such as answering customers, updating records or following up, and you’d rather own that system than rebuild another chain of flows.

Watch for:

it isn’t a drop-in canvas for developer RAG pipelines. For those, start with Langflow or Dify.

Design yours at openlazarus.ai/build

2. Botpress

Best for customer-facing chat agents across websites and messaging channels

A platform to build and deploy LLM agents, built from Workflows made of Nodes, plus Autonomous Nodes.

Price:

Free (100 conversations). Plus $150 a month, billed annually.

Licence:

SDK, CLI and integrations MIT; the platform is hosted

Self-host:

Only the legacy v12 (AGPLv3 or proprietary)

Pick it when:

your Flowise flow was a chatbot talking to customers, and you want channels and conversation tooling.

Watch for:

Autonomous Nodes don’t reach the Conversation End workflow by default.

botpress.com

3. Dify

Best for RAG chatbots and workflows in a hosted, collaborative workspace

Agentic workflows and RAG pipelines, with many models and tools, in one collaborative workspace. Its Chatflow and Workflow app types are the closest match to Flowise’s chatflows and agentflows.

Price:

Free Sandbox (200 message credits, 5 apps). Professional $590 per workspace a year.

Licence:

Dify Open Source License (Apache-2.0 with conditions: no multi-tenant hosting without permission, keep the logo)

Self-host:

Docker Compose (at least 2 CPU, 4 GiB RAM)

Pick it when:

you want the visual RAG and chat experience Flowise gave you, with a team workspace around it.

Watch for:

Chatflows can’t use Triggers.

dify.ai

4. Flow-Wiser

Best for keeping existing Flowise flows running without rebuilding them

An unofficial community continuation fork of Flowise, with dependency and security fixes.

Price:

Free.

Licence:

Apache-2.0

Self-host:

Docker image (dblagbro/flow-wiser)

Pick it when:

you need your flows running as they are while you plan a move.

Watch for:

one maintainer and few users (8 GitHub stars when we checked), so treat it as a bridge, not a home.

github.com/dblagbro/flow-wiser

5. Langflow

Best for the closest visual replacement for Flowise, self-hosted

A visual tool for building and deploying AI agents and workflows. A flow with the Agent component and Chat Input and Output is the closest match to a Flowise chatflow.

Price:

Free and open source. No paid hosted plan is published.

Licence:

MIT

Self-host:

uv or the langflowai/langflow Docker image

Pick it when:

you liked building on a canvas and want the most similar tool, maintained (v1.13.0 shipped October 22, 2026).

Watch for:

it imports Langflow JSON only, and allows one API key per model provider.

langflow.org

6. LangGraph and LangSmith

Best for moving your agents to code on the LangChain stack Flowise was built on

LangGraph builds agents as graphs in code; LangSmith Studio debugs them, and LangSmith Fleet (formerly Agent Builder) is LangChain’s no-code agent builder.

Price:

LangGraph is free. LangSmith Developer is free (5,000 traces, 1 seat); Plus $39 a seat a month.

Licence:

LangGraph MIT; LangSmith commercial

Self-host:

LangGraph yes; LangSmith Fleet self-hosting in beta

Pick it when:

your team can maintain code and wants versioned, testable agents.

Watch for:

Studio’s chat mode works only for graphs whose state extends MessagesState.

langchain.com

7. Microsoft Copilot Studio

Best for agents for companies standardised on Microsoft 365

A graphical, low-code studio for building and managing AI agents and workflows in Microsoft’s cloud.

Price:

$200 a month for 25,000 Copilot Credits, or pay-as-you-go. Free trial.

Licence:

Proprietary

Self-host:

No

Pick it when:

your users, data and admin already live in Microsoft 365.

Watch for:

model response streams time out after 5 minutes, and that can’t be changed.

learn.microsoft.com/en-us/microsoft-copilot-studio/fundamentals-what-is-copilot-studio

8. n8n

Best for agents that run inside your automations across many apps

Workflow automation with native AI: the AI Agent node takes a model, memory and tools, built on LangChain JS.

Price:

No free cloud plan (a free trial); self-hosted is free. Cloud Starter €20 a month, billed annually.

Licence:

Sustainable Use License (internal business use; enterprise features need a licence)

Self-host:

Docker

Pick it when:

your Flowise agent was one step in a bigger automation, triggered by events in other apps.

Watch for:

chain nodes have no memory, and LangSmith tracing works only when self-hosted.

n8n.io

Migration kit: getting out of Flowise cleanly

Nothing imports a Flowise flow, so every move is a rebuild. These steps keep it short and safe.

1. Export every flow. On the canvas, open settings and choose Export Chatflow: you get “<name> Chatflow.json” (or “<name> Agents.json” for an agentflow), JSON with nodes and edges. On Flowise Cloud, the dashboard’s Export button saves all flows and assistants in one file.
2. List your credentials. Exports strip passwords, files and credential IDs, so write down every model key, database and tool each flow uses, and create them again in the new tool.
3. Keep your vector store. If your documents live in Pinecone, Qdrant, Postgres or similar, point the new tool at the same index. Re-embed only if you change the embedding model.
4. Record each flow’s behaviour: the prompts, model, temperature, memory and tools, plus ten real questions with the answers it gives today.
5. Rebuild in the new tool and run the same ten questions. Move on only when the answers hold up.
6. Switch the endpoint or widget, keep the old instance read-only for a few weeks, then shut it down.

Moving to code? Give a coding agent the exported JSON with this prompt: “This is a Flowise flow export (nodes and edges). List each node’s type, settings and connections, then rewrite it as a LangGraph (Python) graph with the same prompts, model settings, memory and tools. Mark every credential as an environment variable and list what you couldn’t map.” Check its output against your ten questions.