FlowiseAI

Overview

FlowiseAI is an open-core visual platform for building and operating AI agents. Henry Heng and Chung Yau Ong founded it in 2023, released it on GitHub that April, and sold the company to Workday in a completed acquisition announced August 14, 2025.^[1][2]

This was an acquisition success, not a failure. A weekend project used open-source distribution to reach tens of thousands of developers, then moved from a drag-and-drop LangChain layer into native orchestration, evaluations, and enterprise controls. Workday bought both technology and community. The standalone service remains live, while a separate Workday Flowise Agent Builder was announced with no current official proof of general availability or revenue.

Founding Story

YC identifies Heng and Ong as active founders. Heng had been a software engineer at Fidelity Investments. In February 2023, the pair were building with LangChain, Hugging Face, and adjacent frameworks and wanted faster prototypes, less repetitive code, and a way to see an entire LLM flow.^[1] Their accessibility metaphor was LEGO: connect a document loader, embeddings, and a vector database while learning how the system worked.

Flowise began as a weekend project. Heng later recalled writing its first line of code on a train from Limerick to Dublin.^[3] The repository appeared publicly in April, YC acceptance followed in June, and a pre-seed plus growth to five employees followed by September.^[4]

The exact circumstances in which Heng and Ong met and their early division of responsibility were not found. The public record is clearer about product purpose than company formation.

Timeline

• February 2023: The founders conceived a visual LLM builder.^[4]
• March 31–April 2023: The public repository was created and Flowise launched as open source.^[5]
• July 5, 2023: YC launched Flowise as a low-code builder for question answering, summarization, and analysis over user data.^[6]
• End of 2023: Flowise reported 19,000 GitHub stars, 9,000 forks, 60,000 npm installs, and 650,000 Docker pulls.^[4]
• August 14, 2025: Workday announced that it had acquired Flowise; later fiscal remarks confirmed the completed transaction.^[2]
• September 16, 2025: Workday announced Flowise Agent Builder inside Workday Build, with planned first-half 2026 availability.^[7]
• 2026: Standalone Flowise remained live; releases 3.1.0–3.1.3 shipped from March through June.^[8]

What They Built

Flowise launched as a drag-and-drop canvas for assembling LLM backends. Users connected models, prompts, document loaders, embeddings, vector stores, memory, and tools into flows, then deployed them without hand-writing all orchestration code.

The product expanded up-stack. Current documentation separates Assistant for beginner-friendly agents, Chatflow for single-agent and retrieval systems, and Agentflow for multi-agent orchestration.^[9] Agentflow V2 introduced native nodes, an execution queue, shared Flow State, branching, loops, resumable human approvals, streaming updates, and MCP-connected tools.^[10]

Flowise now spans more than 100 integrations, RAG, logs, tracing, evaluations, APIs, SDKs, embedded chat, air-gapped deployment, horizontal scaling, RBAC, and SSO. It offers managed cloud and self-hosting.^[9] Most code is Apache 2.0, while the enterprise directory and designated files use a commercial license.^[11]

Market Position

Target Customers

Flowise began with people who wanted to learn and prototype AI applications without extensive code. It expanded toward development teams and enterprises that needed controlled deployment. Current testimonials name InsightSoftware, Thermo Fisher Scientific, BTS, Liverpool Hospital, and Publicis Groupe, but do not establish paid scope or contract value.^[12]

Market Size

No audited customer count, ARR, paid conversion, or market-share figure was found. Workday said Flowise had processed millions of chats and workflows and passed 42,000 GitHub stars at acquisition.^[2] Those metrics show use, not revenue.