FlowProof turns SOPs into continuously tested workflows. Powered by GPT-6 Astra, it executes each step in a browser, detects documentation drift when software changes, recovers from broken paths, verifies outcomes, and proposes evidence-backed updates. Think continuous testing for business operations — with health scores, run history, screenshots, safe-mode controls, and clear documentation patches.
Hey Product Hunt 👋
We built FlowProof around a simple problem: software changes constantly, but the SOPs and runbooks people rely on rarely keep up.
A button gets renamed. A menu moves. A workflow gains an extra step. The documentation still looks perfectly fine — until someone actually tries to use it.
FlowProof treats business processes more like software.
Instead of trusting an SOP because it exists, FlowProof uses GPT-6 Astra to actually execute the workflow, verify each step, detect when the documented path no longer works, recover from changes, and propose an evidence-backed update.
The core loop is:
Run → Detect → Recover → Verify → Patch
One thing we cared about a lot was avoiding the “generic AI agent” approach. FlowProof uses deterministic execution when possible and brings in Astra when the workflow requires semantic understanding, recovery, or judgment.
We also built Safe Mode so potentially destructive actions — sending, publishing, deleting, purchasing, changing permissions — can stop before the final commit.
Our bigger idea is simple:
Software has tests. Infrastructure has monitoring. Business processes should too.
I’d love to hear what workflows you would want FlowProof to test first.