Please feel free to give a feedback. Leave a comments, make an issue on gh or just text me email. Cheers!

Great writeup. The point about AI agents needing structure resonates most with me. I've seen too many teams jump to "let the AI figure it out" and end up with unpredictable behavior in production. The spectrum model you described is important. Not every use case needs a full autonomous agent. Sometimes a structured flow with one AI reasoning step is the right call. The trick is knowing where to draw that line. On the observability side, I'd add that execution traces are essential once you cross into multi-step workflows. If you can't replay what the agent decided at each step,

Ohh nice, one thing I like about this method is that it's easier to maintain. Long prompts tend to become fragile over time, while schemas make the expected output much more explicit.

This really resonates, especially the point about prompts describing behavior but not enforcing it. I have seen a few teams try to solve workflow problems by continuously refining prompts, and it works right up until there's a partial failure or a workflow that needs to resume mid-execution. That's usually when the distinction between "the model should do this" and "the system guarantees this" becomes very obvious. The queue processing example is a good one because the output can look completely reasonable even when the workflow itself has failed. Those are often the hardest i

For me, a tool becomes bookmark-worthy when it consistently saves time. Speed and simplicity are important, but what really makes me return is trust. If I know a tool will work every time without ads getting in the way, forcing signups, or adding unnecessary steps, it earns a permanent place in my bookmarks. I also like tools that help me discover something new rather than just complete a task. For example, I've bookmarked resources that let me explore random countries and learn interesting facts because they're both useful and engaging. The best tools solve a problem quickly while g

This is the part many teams are learning the hard way: AI speeds up writing code, but not understanding it. The bottleneck just moves from coding to review, testing, and ownership.