marvinvista.com

If you're going to really exploit the capabilities of these new tools, you need to be operating at the top of your game. You're not just responsible for writing the code - you're researching approaches, deciding on high-level architecture, writing specifications, defining success criteria, designing agentic loops, planning QA, managing a growing army of weird digital interns who will absolutely cheat if you give them a chance, and spending so much time on code review.

“In AI, we talk about ‘product overhang’, and this is what we discovered with the prototype. Product overhang means that a model is able to do a specific thing, but the product that the AI runs in isn’t built in a way that captures this capability. What I discovered about Claude exploring the filesystem was pure product overhang. The model could already do this, but there wasn’t a product built around this capability!”

…

“Building and testing 5-10 prototype ideas in a day is possible with AI agents. Prototyping used to be so time consuming that if there were two days to prototype, it was lucky to have two distinct prototypes built by the end. But now, agents can build prototypes very quickly, so tests of 5-10 prototypes per day are easily done, as the Claude Code team did.”

…

“Claude generates Markdown which is rendered in the terminal. Before the release, Anthropic engineers complained that nested lists don’t really look that good, and that the spacing between paragraphs was off. The problem was with the markdown renderer. Boris tried all the existing ones, but none looked good inside the terminal. The day before public release, Boris used Claude Code to vibe code a Markdown parser and renderer. And the result was better than anything preceding it, so they shipped it! Boris says he thinks no other terminal has the same Markdown rendering.”

“Upon opening a project or folder, you’re likely to jump into editing files. This means Cursor needs to generate autocomplete suggestions, which the Cursor team calls tab suggestions. A low-latency sync engine powers the ‘tab model’. This generates suggestions that are greyed out and can be accepted by hitting the ‘Tab’ key. The suggestions need to be generated quickly in ideally less than a second. This ‘tab model’ must be as fast as possible, and data transfer as low as possible. There’s always a tradeoff between how much context to send, and the quality of the suggestions: the more relevant context Cursor can send, the better the suggestions. However, sending lots of context can slow down the display of suggestions, so getting this right is one challenge for Cursor’s engineers.”