Flowise
Open source platform for building AI agents and LLM workflows through a visual drag-and-drop editor, available as managed cloud or self-hosted software. The project is being sunset, with end of life on 31 August 2026.
What is Flowise?
Flowise is an open source generative AI development platform for building AI agents and LLM workflows. Rather than writing orchestration code by hand, you assemble a flow on a visual canvas where every integration is a node, connected much like building blocks.
The platform ships three visual builders. Assistant is the most approachable: it produces a chat assistant that follows instructions, calls tools when needed, and retrieves knowledge from uploaded files through RAG. Chatflow targets single-agent systems and chatbots, and unlocks more advanced techniques such as Graph RAG, rerankers and custom retrievers. Agentflow is a superset of both, used for multi-agent systems and complex workflow orchestration.
Around these builders sits a complete toolchain: execution tracing and analytics, evaluations built on datasets and evaluators, human-in-the-loop review of agent tasks, document stores and upsertion, response streaming, file uploads, variables and team workspaces. Orchestration supports both open source and proprietary models, expressions, custom code, and branching, looping and routing logic. Data ingestion reaches more than 100 sources, tools, vector databases and memory backends. Monitoring covers execution logs, visual debugging and external log streaming, with Prometheus and OpenTelemetry among the supported targets.
Flows are meant to leave the canvas. A public REST API exposes assistants, chatflows, predictions, document stores, tools, variables and vector upsert; TypeScript and Python SDKs and a command line interface complete the picture, alongside a customisable embedded chat widget. Model Context Protocol nodes act as client or server, with tool listing, SSE and authentication. Enterprise controls include RBAC, SSO, encrypted credentials, secret managers, rate limiting and restricted domains, while deployments scale horizontally through a message queue and workers, air-gapped installations included.
The software is Apache 2.0 licensed and installs locally with two npm commands or runs under Docker; a managed cloud is offered alongside. Priceline, Deloitte, Publicis, Globe Telecom, QMIC and UneeQ appear as customers on the site, and the repository has passed 42,000 GitHub stars.
Flowise is now being wound down. Feature development ceased on 29 July 2026, the repository was archived on 10 August 2026, and end of life falls on 31 August 2026. The code stays public and forkable.
What it does
- Build AI agents visually, by dragging and connecting nodes instead of writing orchestration code
- Orchestrate multi-agent systems with branching, looping and routing logic
- Ground answers in your own documents through retrieval-augmented generation over 100+ sources
- Insert human review steps inside an agent's feedback loop before actions are carried out
- Trace, log and visually debug every execution, with Prometheus and OpenTelemetry support
- Expose any flow through a REST API, TypeScript or Python SDK, or an embedded chat widget
- Run the whole platform on your own infrastructure, including fully air-gapped installations
When to use Flowise / When not to
A quick filter to help you decide if Flowise is the right fit.
When to use Flowise
- Engineering teams prototyping LLM agents without writing the whole orchestration layer themselves
- Organisations that need on-premises or air-gapped deployment and full control over where data sits
- Teams willing to fork and maintain the Apache 2.0 code internally now that the project is winding down
- Product and support teams embedding a retrieval-based chat assistant into an existing application
- Consultants and independent developers who need to stand up agent prototypes quickly for clients
When not to use Flowise
- Teams starting a long-term project, since feature development stopped on 29 July 2026
- Non-technical users who want self-hosting but have no server, database or upgrade experience
- Anyone looking for a mobile app, as neither an iOS nor an Android application exists
- Buyers who require published security certifications such as SOC 2 or ISO 27001
- Organisations that need a vendor support commitment beyond the 31 August 2026 end of life
How to use Flowise
A typical end-to-end flow, from setup to results.
- Decide between the managed cloud and self-hosting; the documentation recommends the cloud to anyone not comfortable running servers, backing up databases and applying updates
- For the fastest start, create an account at cloud.flowiseai.com and begin on the permanent free plan
- To run it locally, check that NodeJS v18.15.0 or v20 and above is installed, then run npm install -g flowise
- Start the instance with npx flowise start and open http://localhost:3000 in a browser
- Alternatively, clone the repository, copy .env.example to .env in the docker folder and run docker compose up -d
- Pick a builder to match the job: Assistant for a guided chat assistant, Chatflow for a single agent, Agentflow for multi-agent orchestration
- Drag the nodes you need onto the canvas, connect them, and attach your model, tools, memory and vector store credentials
- Load your documents into a document store and upsert them so the flow can retrieve from your own knowledge base
- Test the flow, inspect the execution traces, and set up evaluations with datasets and evaluators before going further
- Ship it by calling the prediction endpoint from your application, installing the SDK, or pasting the embedded chat widget into your site
Pros & Cons
Pros
- Apache 2.0 licensing means the code stays usable and forkable even after the shutdown
- Visual building makes agent development accessible without writing the whole stack by hand
- Covers the full lifecycle: prototype, evaluate, observe and run in production
- Self-hosting and air-gapped deployment give complete control over where data lives
- In self-hosted mode the vendor collects no metrics and no data whatsoever
- Broad ecosystem reach, with LangChain, LlamaIndex, LiteLLM and 100+ models and vector stores
- A permanent free cloud plan and an active community of more than 42,000 GitHub stars
Cons
- The product is end of life: development has stopped, the repository is archived and support ends 31 August 2026
- npm packages and Docker images are marked deprecated, so no security fixes should be expected
- No security certification is published, neither SOC 2 nor ISO 27001
- No Data Processing Agreement is offered and no formal subprocessor list is published
- The vendor never states its position on training AI models with customer data
- Cloud hosting sits only in the US East 1 region, with no EU data residency option
- Self-hosting demands genuine server skills, and parts of the documentation are a year old
Pricing & Plans
A permanent free plan is available at 0 USD per month, covering 2 flows and assistants, 100 predictions per month and 5 MB of storage. The lowest paid entry point is the Starter plan at 35.00 USD per month, advertised with the first month free. Prices exclude applicable taxes and duties, and Stripe acts as merchant of record. Self-hosting the Apache 2.0 software carries no licence fee, leaving only the cost of the infrastructure it runs on.
- 2 flows and assistants
- 100 predictions per month
- 5 MB storage
- evaluations and metrics
- custom embedded chatbot branding
- community support
- everything in Free
- unlimited flows and assistants
- 10
- 000 predictions per month
- 1 GB storage
- community support
- first month free
- everything in Starter
- 50
- 000 predictions per month
- 10 GB storage
- unlimited workspaces
- 5 users then 15 USD per additional user per month
- admin roles and permissions
- priority support
- licence key required (FLOWISE_EE_LICENSE_KEY)
- arranged by contacting hello@flowiseai.com
- Self-hosted — no licence fee under Apache 2.0
- infrastructure costs only
Data, GDPR & hosting
A consolidated view of how Flowise handles your data.
GDPR overview
The privacy policy, effective 3 June 2024, addresses the GDPR concretely rather than in passing. It sets out four legal bases for processing: contractual necessity, legitimate interests, legal obligations and consent. It lists data subject rights in full, including access, rectification, erasure, restriction, objection, withdrawal of consent, portability and the right to complain to a supervisory authority, with the EDPB named. Requests go to privacy@flowiseai.com. Transfers outside the user's region are covered by Standard Contractual Clauses, and consent is collected through a cookie banner where the law requires it. Three gaps are worth noting: no Article 27 EU representative is designated, no data protection officer is named, and no Data Processing Agreement is published or offered.
Who owns the data?
Users keep ownership of their content. The terms state that you retain ownership of your User Content and grant FlowiseAI only a non-exclusive, worldwide, royalty-free licence to use, host and process it, solely to operate and improve the Services. FlowiseAI explicitly claims no ownership over the workflows or models you build and undertakes not to use them publicly without your consent. The vendor does reserve the right to use anonymised and aggregated usage data to improve the platform. In self-hosted deployments the question is largely moot: the privacy policy states that no metrics and no data are collected at all.
Reuse rights
Nothing in the terms restricts what users may do with the workflows, configurations and outputs they create: ownership stays with the user, and no separate permission is required to reuse, modify, commercialise or move that material elsewhere. The licence granted to FlowiseAI runs in the other direction and is narrow, covering only hosting and processing needed to operate and improve the Services. Website content, by contrast, remains the property of FlowiseAI or its licensors and may be downloaded and shared for internal, non-commercial purposes only. The platform software itself is published under Apache 2.0, so the code can be forked, modified and redistributed freely, which the team has actively encouraged since announcing the shutdown.
Data retention & training
Hosting summary
For Flowise Cloud, the privacy policy states plainly that user data is stored in the US East 1 region, placing it under United States jurisdiction. No European or other regional hosting option is offered, and no data residency choice is presented at sign-up. Where data is transferred outside a user's own region, the vendor says it applies safeguards such as Standard Contractual Clauses. Third parties named in the policy as handling data include Stripe for payments, PostHog for product analytics and Google Analytics for website usage, alongside unnamed hosting providers; this list is given by way of example rather than as a formal subprocessor register. Self-hosted deployments change the picture entirely: the infrastructure is yours, and the privacy policy confirms that no metrics and no data are collected from self-hosted installations. Note that the flowiseai.com marketing site itself resolves to an Amazon Web Services address in Frankfurt, Germany, which reflects where the website is served and not where customer data is kept.
Things to keep in mind
Risks and trade-offs to weigh before adopting Flowise.
- The end of life on 31 August 2026 removes the core team from Discord and GitHub, so no vendor will answer when something breaks
- Deprecated npm packages and Docker images mean security vulnerabilities in the stack will not be patched upstream
- Adopting an unmaintained platform quietly transfers the maintenance burden to your own team, a cost that is easy to underestimate at the moment of choosing
- Cloud data sits only in the United States, which is a real constraint for organisations with EU residency obligations
- The absence of a Data Processing Agreement, a subprocessor list and any security certification leaves procurement and compliance teams without the documents they normally require
- Because the vendor never states whether customer data feeds model training, teams may assume an answer that was never given
- Visual agent builders make it easy to ship a flow that looks convincing without understanding what it does, so treat human-in-the-loop review and evaluations as necessary rather than optional
Setup & Integrations
Technical difficulty
It depends entirely on the route. The cloud requires no technical skill: create an account and start on the free plan. A local install is two npm commands, provided NodeJS v18.15.0 or v20 and above is present, with the interface on localhost:3000. Docker Compose adds a clone and an environment file. Running it properly in production is a different matter, involving a database, environment variables, SSO, rate limiting, a message queue and workers. The documentation warns directly that self-hosting requires the skill to set up an instance, back up the database and maintain updates.
Deployment
Integrations
Supported languages
Behind Flowise
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Is Flowise still being maintained?
What happens to the code now that the project is closing?
Is there a free version?
How much does the cheapest paid plan cost?
Can I run Flowise on my own infrastructure?
Does Flowise offer an API?
Where is cloud data hosted?
Is my data used to train AI models?
What is the minimum age to use Flowise?
Who is behind Flowise?
Should you pick Flowise?
Flowise built a genuinely capable product: three visual builders that scale from a guided chat assistant to full multi-agent orchestration, retrieval over more than a hundred data sources, execution tracing, evaluations, human-in-the-loop review, and enterprise controls such as RBAC, SSO and air-gapped deployment. More than 42,000 GitHub stars and customers of the size of Deloitte, Publicis and Priceline suggest the adoption was real rather than aspirational.
That makes the current situation unusual. The team announced the wind-down on 29 July 2026, archived the repository on 10 August 2026, and set end of life for 31 August 2026, reasoning publicly that rigid low-code workflows hit their limits as coding agents grow more capable. Feature work has stopped, pull requests are locked, and npm packages and Docker images are deprecated. FlowiseAI Inc. was acquired by Workday in August 2025, and the service terms now point to Workday's own terms.
What remains is the code, under Apache 2.0, and the team's explicit invitation to fork it. That is a meaningful asset for an organisation with the engineering capacity to carry it, particularly one already running Flowise in production and needing a controlled path forward. It is not a reasonable foundation for a new project with a multi-year horizon, and the gaps that were tolerable in a maintained product become harder to accept in an unmaintained one: no published security certification, no Data Processing Agreement, no subprocessor list, no stated position on model training, and cloud hosting confined to a single United States region.
Evaluate Flowise today as source code you may adopt, not as a service you can buy. If you can maintain it internally, the architecture and documentation are worth inheriting. If you need a supported vendor relationship, look elsewhere.
- Choosing a selection results in a full page refresh.
- Opens in a new window.