This project seems like a good idea that didn't have enough of a moat. I'd suggest trying to narrow your target customer from "engineers that want to manage agents on their phone" to "people trying to do some particular kind of task," so you can bake in more value add and automation. You're not going to beat the labs on general tools, but they're not going to be willing to narrow their target customer, so you'll always be able to win at the margins.