GoKawiilThere's a moment every product team recognizes. Someone on the roadmap wants a visual workflow editor – drag-and-drop nodes, connected logic, the ability for users to build their own automations. The conversation turns to build vs. buy, and someone says:
"We can just use React Flow. How hard can it be?"
It's a reasonable question. React Flow is open source, well-documented, and has 35,000+ GitHub stars. The answer, though, is that React Flow gives you a canvas and some node primitives. Everything else – the edge routing, the auto-layout, the node configuration panels, the validation, the execution visualization, the design system, the performance tuning – you build from scratch. And that's where estimates go wrong.
This article breaks down what building a workflow editor in-house actually costs, why the initial estimate is almost always understated, and where the compounding maintenance burden tends to catch teams off guard.
Why the estimate is ALWAYS low
The initial scoping of a workflow editor typically looks deceptively clean: set up React Flow, define a few node types, add a sidebar, wire up save/load. A senior developer might quote a maximum of a few weeks.
That estimate captures the happy path. It doesn't capture the problems that only emerge once you start building, or once users start using what you've built.
Edge routing
Edges that avoid nodes and overlap cleanly are one of the hardest problems in workflow editor UI. React Flow's default edges take the shortest path. In a dense workflow, they cross through nodes, overlap each other, and become unreadable. Building obstacle-avoiding, collision-free routing from scratch typically takes two to four weeks of dedicated engineering time… and it's almost never in the original estimate.
Auto-layout
... continue reading