Flowise
Last reviewed
Jun 4, 2026
Sources
19 citations
Review status
Source-backed
Revision
v1 · 2,028 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 4, 2026
Sources
19 citations
Review status
Source-backed
Revision
v1 · 2,028 words
Add missing citations, update stale details, or suggest a clearer explanation.
Flowise is an open-source, low-code platform for building large language model applications, chatbots, and AI agents by dragging and dropping nodes onto a visual canvas. Developed by FlowiseAI Inc., the project was open-sourced on GitHub in early 2023 and grew into one of the most widely used visual builders in the generative AI tooling space, accumulating tens of thousands of GitHub stars. The company was founded by Henry Heng and Chung Yau Ong, went through Y Combinator's Summer 2023 batch, and in August 2025 was acquired by the enterprise software company Workday. Flowise lets users assemble retrieval-augmented generation pipelines, multi-agent workflows, and conversational assistants without writing code, while still exposing an API and SDKs for developers who want to embed or extend the flows they build.
Flowise was created by Henry Heng (also written Zhenjing "Henry" Heng), who serves as chief executive officer, together with co-founder Chung Yau Ong. Before starting the company, Heng worked as a software engineer at Fidelity Investments. The founders' stated goal was to make it easy for people without a coding background to learn and build AI applications. The project was first released publicly on GitHub in early 2023, around the time the post-ChatGPT wave of LLM tooling was taking off, and the codebase was published under the permissive Apache License 2.0.
The early product was framed as "an open source drag and drop tool to build your customized LLM flow." Users could connect components such as document loaders, embedding models, and vector databases on a canvas to assemble a backend for question-answering, summarization, and analysis tasks, an approach the founders compared to building with LEGO bricks. From the start the tool leaned heavily on the LangChain framework, and it later added support for LlamaIndex, Hugging Face, and other libraries.
FlowiseAI participated in Y Combinator's Summer 2023 (S23) batch with a team of five, and the startup's YC group partner was Harj Taggar. As is standard for accelerator participants, the company received Y Combinator's investment under the accelerator's standard deal. Public startup databases including Tracxn, Seedtable, and Crunchbase list FlowiseAI as having raised about $500,000 in total, with Y Combinator recorded as the sole investor, which matches the YC standard deal of $500,000. There is no record of a separately negotiated venture round before the acquisition.
At its YC launch, Flowise had just crossed 10,000 GitHub stars and had a Discord community of more than 2,000 members. The project's popularity grew quickly thereafter: by mid-January 2025 it was widely cited at roughly 30,000 stars, and by the time of its acquisition in August 2025 the figure had passed 42,000.
On August 14, 2025, Workday, a cloud provider of human capital management and financial software, announced that it had acquired Flowise. Financial terms were not disclosed, and the companies did not state a closing date in the announcement. In its press materials Workday described Flowise as a low-code platform for building AI agents, "from simple chatbots to complex automated workflows," and cited the project's traction: more than 42,000 GitHub stars, millions of chats and workflows processed, and adoption across consulting, finance, healthcare, and customer support.
Workday positioned the deal as a way to give its own customers and partners an "industry-leading agent builder" for designing, launching, and managing AI agents for HR and finance. Peter Bailis, Workday's chief technology officer, said the acquisition would let customers "build and deploy their own AI agents on Workday" while investing in the open-source foundation. Henry Heng said that by joining Workday the team could accelerate its vision of "enabling anyone to create powerful AI agents, without needing deep technical expertise." Workday serves more than 11,000 customers globally, including over 60 percent of the Fortune 500, which it presented as the distribution behind the deal.
Both Workday and the Flowise team stated that the open-source project would continue. In a blog post marking the deal, the Flowise team wrote that "Flowise isn't going anywhere, we're doubling down," and reaffirmed its commitment to the open-source community. After the acquisition Flowise continued shipping releases, and the GitHub repository reported a higher star count over time, reaching the low-to-mid 50,000s by 2026 (alongside roughly 24,000 forks) with the 3.x release series and a 3.1.x version line.
Flowise is a self-hostable, web-based application (built primarily in TypeScript and JavaScript on Node.js) that presents a visual editor where each node represents a building block, such as a model, a memory store, a retriever, a tool, or an agent. Connecting nodes defines the data flow of an application. The platform supports branching, looping, and routing logic so that flows can include conditional paths rather than only linear pipelines.
Flowise organizes its builders into a few distinct surfaces, which roughly map to increasing complexity:
| Builder | Purpose |
|---|---|
| Assistant | The most beginner-friendly surface for creating a single AI assistant with instruction-following, tool use, and knowledge retrieval, configured largely through forms rather than a full graph. |
| Chatflow | A canvas for single-agent systems, chatbots, and simpler LLM flows, including RAG techniques such as Graph RAG and reranking. |
| Agentflow | A superset of Chatflow and Assistant aimed at multi-agent systems and more complex workflow orchestration, where multiple coordinated agents and tools can be chained together. |
This split lets a non-technical user start with an Assistant, move to a Chatflow for a custom chatbot, and graduate to an Agentflow when a use case requires several agents working together with human-in-the-loop checkpoints.
Flowise is model-agnostic. It advertises support for more than 100 integrations across LLMs, embedding models, vector stores, and tools. Users can connect commercial model providers such as OpenAI and Anthropic, as well as locally hosted or open-weight models, which makes it possible to keep data and inference on-premises. On the data side, Flowise ships nodes for document loaders (text, PDF, DOC, SQL, and other sources), embeddings, and a range of vector databases, which together let users build RAG pipelines that ground model responses in their own documents. Because the project is built on top of the LangChain and LlamaIndex ecosystems, many components map onto abstractions from those frameworks.
Beyond retrieval, Flowise supports tools and agents that can take actions, for example calling an external API to retrieve information and then reasoning over the result. The platform also includes data transforms, filters, and aggregation nodes, plus datasets and evaluators for testing flows.
Once a flow is built, Flowise exposes it in several ways. Each flow can be called through a REST API, and the project provides JavaScript/TypeScript and Python SDKs as well as a command-line interface. For end-user deployment, Flowise generates a customizable embeddable chat widget that can be dropped into a website or web application. This combination, fast visual assembly plus an API and an embeddable widget, is one of the main reasons the tool became popular for building chatbots and internal tools quickly.
For operations, Flowise offers full execution traces and visual debugging, with observability that can export to Prometheus and OpenTelemetry. Enterprise-oriented capabilities include role-based access control, single sign-on, encrypted credential storage, and air-gapped (fully self-hosted) deployment.
Flowise is distributed in two main ways: a free, self-hosted open-source edition and a managed Flowise Cloud service, with an additional enterprise tier.
The Community Edition source code is licensed under Apache 2.0, which allows free use, modification, and redistribution, including for commercial purposes. The project also includes enterprise functionality whose source is governed by separate commercial terms rather than the Apache 2.0 license, a split the maintainers have clarified in the repository. The self-hosted edition can be installed quickly via npm (for example, installing the flowise package globally and running npx flowise start) or deployed with Docker, and it runs on common cloud platforms such as AWS, Azure, Google Cloud, DigitalOcean, and Railway.
Flowise Cloud is the hosted option for users who do not want to manage infrastructure. Public pricing for the cloud service has been listed approximately as follows (subject to change):
| Plan | Price | Notable limits |
|---|---|---|
| Free | $0 / month | 2 flows, 100 predictions per month, 5 MB storage |
| Starter | $35 / month | Unlimited flows, 10,000 predictions per month, 1 GB storage |
| Pro | $65 / month | 50,000 predictions per month, 10 GB storage, unlimited workspaces and user management |
| Enterprise | Contact sales | On-premise or cloud deployment, SSO, RBAC, and additional governance |
Flowise reports usage by a range of large organizations. Logos shown on its site and in its YC profile include AWS, Accenture, Deloitte, Publicis Groupe, Priceline, Globe Telecom, Deriv, Thermo Fisher, and others, spanning consulting, finance, healthcare, telecommunications, and customer support. Workday's acquisition announcement reiterated that the platform processes millions of chats and workflows. The tool is also a frequent subject of tutorials and deployment guides (on platforms such as Railway, AWS, and Docker), reflecting its strong open-source following.
Flowise competes with a cluster of visual or low-code platforms for building LLM applications and agents, most often compared to Langflow, Dify, and n8n.
| Tool | Positioning | Notes |
|---|---|---|
| Flowise | Lightweight, drag-and-drop builder emphasizing chatbots, RAG, and embeddable widgets | Built on LangChain and LlamaIndex; praised for getting to a working prototype fastest, with a smaller footprint and fewer moving parts. |
| Langflow | Visual companion to LangChain, strong on RAG and Python flexibility | Flows compile to LangChain code; favored by engineering teams already invested in LangChain. |
| Dify | Full LLM-application platform with backend, admin UI, and API gateway | Strong on production features such as queue management, multi-tenant workspaces, and role-based access; often recommended for teams shipping LLM-powered SaaS. |
| n8n | General-purpose workflow automation tool with AI nodes added | Broader automation use cases beyond AI; can feel heavier for users who only need simple AI workflows. |
Comparisons commonly characterize Flowise as the "lighter, faster" choice that trades some advanced production capabilities (deeper observability, full multi-tenancy, robust versioning) for simplicity and speed of initial deployment, while Dify leans toward production infrastructure and Langflow toward developer workflows. Open-source star counts among these projects have shifted over time; by early 2025, comparison write-ups put Dify in the high 50,000s, Langflow around the low 40,000s, and Flowise around 30,000, though all three have grown since.