My AI built a flight tool, then talked me out of rebuilding the hard part
I had my agent build a tool to find the cheapest flights. Then I asked the obvious founder question: can we just make the whole thing ourselves and drop the dependency? The answer it gave me was the valuable part.
Build what is yours, borrow what is a war
The flight tool works by fanning a search across hundreds of sources and returning the best price. My instinct was to own all of it. My agent laid out why that was a bad trade. Replicating the price-scraping fleet means an endless cat-and-mouse fight against sites that do not want to be scraped. That is somebody else's whole company, not a feature we should rebuild.
The smarter move it proposed was to rebuild the tool on an official data source, the proper front door that airlines and agencies actually use, so we have something reliable and owned without fighting a war we would lose. Same tool from my side, sturdier foundation underneath.
Knowing when not to build is the skill
Anyone can be told to build a thing. A good partner tells you when building it yourself is a trap. The honest trade-off, reliability and ownership versus chasing the last few dollars of savings, is exactly the conversation a founder needs.
We build a lot. We are just as useful when we tell you not to.
Sometimes the best build is the one we talk you out of. That honesty is part of the deal.
← Back to all posts