Started October 2026
Things I reckon will happen, written down with a date so I can look back and see how they aged.
Happy to be proven wrong.
19 Oct 2026
When I come back to a piece of work I only want to know where we're at and what's needed next. Scrolling back through a thread to work that out is tiring, especially with a few agents on the go. A board shows it at a glance, which is partly why I built Otto.
Tools are slowly heading this way. I'm a bit surprised it isn't happening faster.
19 Oct 2026
An agent on my laptop picks up everything about my laptop: odd versions, old environment variables, a database I seeded by hand. It can't see any of that, so it guesses. Another black hole.
In the cloud every environment is the same, can be thrown away, and you can run as many as you need. It's the next step on my AI engineering journey.
19 Oct 2026
I have to pick a model before I know how hard the task is. The provider is in a much better position to make that call, and models already decide how long to think.
Either it becomes automatic at their end, or one model gets efficient enough that there's nothing left to choose.
19 Oct 2026
Most of the noise is engineers building tools for other engineers, where everything moves at the speed of a merge. Building for the real world doesn't work like that.
People need time to trust something new and change how they work, and that doesn't speed up just because the code got written faster. It's just not as exciting to post about as merging 2,000 PRs a month.