Sharing my progress on a build—would love some feedback!
Hey guys, sharing my progress on a build and would love to get some feedback. I got sick of AI agents hallucinating complex pricing math and ruining quotes, and the Google Sheets webhook route was just too clunky. So, I built an external "math brain" to handle it. The workflow is simple: 1. The AI just handles the chat and collects the variables. 2. It fires a webhook to a visual logic engine I built. 3. The engine runs the strict math (conditional fees, distance, bundles) and returns a clean, hallucination-free JSON payload back to the agent. Here is a 3-minute Loom showing the backend builder and the API handoff: https://www.loom.com/share/edc0ecd308124007b2f84bf777f9283d Would love your brutal feedback on this architecture. Where do you guys see this breaking in your own workflows?