I love that it ties signals directly to your own product and stakeholders, so the alerts actually connect to why someone might buy (instead of generic, "throw stuff at the wall and see what sticks" approach). What really stood out to me though was the MCP server. I set it up myself in Claude Code and ran the whole pipeline end-to-end (config, signals, discovery) straight from the command line without a dashboard needed. As a developer, having a product I can drive entirely through an AI agent rather than clicking through a UI is to me a real differentiator. Most tools in this space don't offer that yet. Review collected by and hosted on G2.com.

I did run into a few things while testing that I think are worth ironing out. Signal generation doesn't seem to check against signals I already had, so I ended up with several near-duplicates. I also hit a run where discovery skipped almost everything with no reason given in the response - I eventually figured out why, but only by comparing multiple runs myself, which shouldn't be necessary. And my account list ended up populated with a mix that didn't quite match the personas I had set up, which quietly killed a lot of my results until I caught it. None of it felt broken though, and since it is still in early states these are the kinds of things I expect Trayo to get sanded down as the product matures. Review collected by and hosted on G2.com.