Flowise Is Shutting Down a Year After Workday Bought It. Coding Agents Ate the Business That Acquired It.
Flowise Is Shutting Down a Year After Workday Bought It. Coding Agents Ate the Business That Acquired It.
Workday acquired Flowise, the popular open-source low-code platform for building AI agents, on August 14, 2025. On July 29, 2026, eleven and a half months later, Flowise announced it's winding down. The stated reason, straight from the company's own sunset page, isn't a funding problem or a legal issue. It's that developers moved to coding agents like Claude Code, and a drag-and-drop flow builder "quickly hits the limit when it comes to complexity." The source code stays on GitHub under Apache-2.0, so it's forkable. The product and the company behind it are done.
If any part of your stack, a client deliverable, or an internal tool runs on Flowise or a tool like it, this is worth reading past the headline.
The timeline, both ends of it from primary sources
Workday's own August 14, 2025 press release framed the acquisition as bringing "powerful AI agent builder capabilities to the Workday platform." Flowise had real traction going into that deal: it launched about two years earlier with Y Combinator backing, co-founded by Henry Heng and Chung Yau Ong, and had built a large open-source community around its visual, low-code approach to chaining LLM calls, tools, and data sources into working agents. The financial terms of the acquisition were never disclosed.
Flowise's own sunset page gives the reason plainly: over the months leading up to the announcement, the team noticed a shift in how people build, with developers increasingly relying on coding agents such as Claude Code to handle complex tasks that a rigid, low-code workflow approach can't keep up with. That's the company's own words for why it's ending, not a leak or an analyst's guess. The page also lays out a specific wind-down schedule: code freeze and the public announcement on July 29, the GitHub repository moving to public archive status on August 10, and the core team's presence in Discord and GitHub ending on August 31, described as the project's official end of life.
Why "acquired, then shut down a year later" is worth sitting with
This isn't a story about a startup running out of money. Flowise had a real acquirer with real resources behind it. The shutdown reads less like a failure to execute and more like a specific bet, that low-code AI-agent building has a durable future distinct from coding agents themselves, that the company backing it apparently stopped believing within twelve months. An acquisition is usually read as validation. Here, the acquirer's own product line didn't survive a year past the deal closing, which is a much sharper signal about the category than a startup simply failing to raise its next round would have been.
What this actually means if you built on Flowise
Apache-2.0 licensing means the code isn't disappearing, and neither are any flows you've already built on top of it. You can fork the repository, self-host it, and keep the existing infrastructure running indefinitely if you choose to. What does disappear is community support at the scale it existed before, hosted infrastructure if you were using it, and any expectation that the tool's roadmap moves forward. If you've got a client deliverable, an internal automation, or a product feature built on Flowise, the useful move is auditing what's actually running on it now and deciding, on your own timeline, whether to maintain a self-hosted fork or migrate, rather than finding out something's broken later with no one left to ask.
The bigger claim, and why I'd hold it more loosely
It's tempting to read this as proof that the entire low-code AI-agent-builder category is dying, replaced wholesale by coding agents. I don't think the evidence supports that read as cleanly as the headline suggests. n8n, Zapier's AI features, and several other low-code and no-code automation platforms are still shipping, still growing, and serving a real audience: people who build useful automations and don't write code, and were never going to sit down with Claude Code in a terminal regardless of how good it gets. Flowise's own explanation, that its users specifically moved toward coding agents, may say more about who was actually using Flowise (a technical audience comfortable enough with LLM tooling to appreciate a visual workflow builder, and just as comfortable graduating past it) than about the fate of low-code AI tooling as a category.
The honest counter
I could be wrong about how narrow this is. If coding agents keep getting better at handling ambiguity and ship with better default guardrails, the argument for a visual, constrained workflow builder gets weaker across more of the market, not just for Flowise's specific power-user audience. A year from now, this could look like the first domino in a broader shift rather than one company's specific and narrower story. I don't think that's the likelier read today, but it's the one I'd flag as the way this piece could age badly.
What I'd actually do
Before you adopt any no-code or low-code AI tool for something that matters, go read the vendor's own public communications, not just the pricing page, and ask whether they talk about the tool as durable infrastructure or as a stepping-stone to something else. Flowise's team told its users the answer in its shutdown post. The honest version of that answer would have been more useful to know before building on it than after. If you've got something running on Flowise today, don't panic: Apache-2.0 means you have time and a real path to keep it running. Just don't mistake "the code still works" for "the vendor still has your back," because those stopped being the same thing the day the sunset page went up.
Author
Lukas
@lukcombinator