August 8, 2026/3 min read
Maybe I’m just too cynical but it’s really starting to feel like more or less all of this new “AI”/agent software is vapourware. I’ve been trying to take this whole “agentic development” thing seriously and there is just so much friction in this ecosystem it’s absolutely insane. It’s very clear that all this stuff is built by engineers who a) have free access to unlimited tokens and b) have no systematic review or QA process for usability testing or dogfooding their products except on the most straightforward of happy paths.
The hard part of software engineering was never writing the code. Now that the cost of writing code has collapsed almost down to the cost of LLM inference, code is ubiquitous. Everybody has a vibe coded startup and many are great ideas but the execution seems to lack any regard whatsoever for the user experience.
I tried ONA recently, for example. In principle it’s a great idea. I’d love to be able to make vague requests for improvements to my apps and have a fully automated end to end process for refining those wishes into clear requirements, reconciling them with all the other open feature requests and planned work, scoping them down to implementable features, then running them through design, testing, implementation, QA, and deployment one by one. Of course I want that, my life as a software engineer would be so much easier. This is roughly what this platform says it is capable of. I think this is a very cool idea so I signed up.
My very first encounter with the app was that I was unable to login on their desktop version at all. Auth is hard, so I can empathize and forgive this. So I tried signing in on the web. It took many clicks and redirects but eventually I was in! I was optimistic, but once I had it wired up to one of my projects, instead of making magical hands-off progress on my todo list it spent nearly my entire $20 worth of “ona compute units”, whatever those are, thrashing and trying to get a hold of the todos from linear just so it could pick one to start.
This only adds more friction between me and my projects, which is literally the opposite of what I want when I pay for developer tooling. I’m not inclined to pay more money to see if it can do a better job next time trying to actually build anything, because I’m very skeptical that it can, and building things isn’t what I want AI to do for me anyway. The thing is, I like writing code. That was the fun part of the job until very recently. I want AI to take on all those other annoying parts of the software development lifecycle that were a distraction from writing code. But so far most of what they do is make more annoying work for me. More integrations to debug, more auth tokens to refresh, more bills to keep track of, more meandering and drawn out descriptions of non-bugs with bogus fix suggestions and misinformed “root cause” analyses. It’s tiresome.
The worst part is that the marketing is so overblown. These companies are themselves the best argument against their own sales pitch. If agentic development actually worked the way any of them say it does, then the people selling it would be the primary beneficiaries. They would be shipping the most polished and useful software on the market. Instead I could barely get past the login screen.
The thing is I think ONA is a good idea. I am evidently willing to pay for this kind of tooling. But I want a version that works without lighting twenty dollar bills on fire.