Ask HN: What single AI tool/technique 10x'd your productivity last year?

Posted by laxmena 3 days ago

As someone tinkering with AI agents daily, I’ve seen massive shifts since mid-2025.

What’s the single AI tool, feature, or technique that gave you the biggest productivity leap in the last year? (Cursor Composer? Claude 4.5 projects? A custom RAG setup? Vibe-coding with o1?)

Comments

Comment by PaulHoule 3 days ago

None of the above. I use Junie to help code and think I am writing better code if not writing it faster.

Copilot also helps me transform into a fox and I have all kinds of KPI for that (e.g. got petted) but as-a-fox or otherwise I reject that there is anything 10x about it.

If you think you are 10x more productive coding before I implore you to look at it from the viewpoint of your team. I've been the guy who picks up projects that were "80% done" and had to do the last 80%. What gets me is the inconsistency, the times it does something better than I would do myself instantly vs the times I want it touch 2 files and it modified 35 and I can't tell what the hell it was thinking and frankly don't care.

Comment by MattGaiser 3 days ago

Let Claude have access to everything, have it run a local environment as close to production as possible (Docker with the real dependencies, not some locally installed finnicky Postgres), and tell it to thoroughly test all its work and make sure it is set up to do that.

Paste in ticket, get 95% of the way to the end.

I will add the caveat that this is only with the code. The overhead of communication has not improved at all beyond pasting code review comments into Claude to get replies.

What I really need is something to listen in at meetings for my name, summarize what was asked, and suggest an answer based on the codebase.

Comment by SOLAR_FIELDS 3 days ago

The live transcription tools like MacWhispr that run local transcription can output to a text. Probably a weekend project to do what you’re saying, have a model read the text in realtime somehow and trigger a code search when it hears your name

Comment by tokkyokky 2 days ago

Claude Code for me. The optimization cycle is ridiculously fast – they ship improvements based on user feedback way quicker than I expected. And honestly, Opus 4.5 is smart enough out of the box that I haven't had to mess with configs or prompts much at all.

Comment by porksoda 2 days ago

Are they still arbitrarily changing the keyboard shortcuts around without any option to customise?

Made me so mad.

Comment by oriel 1 day ago

This is one of the things that made me ditch Cursor: when they're constantly shifting around keybindings, overriding long-held conventions, and changing UI, etc etc; especially without opt-out (or opt-in for that matter).

Tie that in with a release cycle that force restarts the IDE, its a recipe for flow-breaking frustration that makes a tool untrustable. When its untrustable, it gets removed.

Now I don't let my tools auto-update. It was an experiment to allow the auto-updates to begin with, because new tech paradigms, yet its clear the end user still isn't accounted for properly.

IMO because the product roadmap (such as it might ever be) is usually only influenced by poorly targeted analytics (more is not better), and a small minority of vocal users; leaving the rest of us in the lurch.

Comment by tokkyokky 2 days ago

Ah, you're right..

Comment by RicoElectrico 3 days ago

I just use Gemini thinking model until the free quota runs out. If you know what you need, it's excellent at coding, makes very little mistakes.

Comment by Octokat 3 days ago

I use Cursor with a max subscription through work and using Gemini in multi-model mode alongside Opus 4.5 and GPT 5.2 Codex, Gemini's been the worst performer. Not saying it's a bad model just that the other 2 do a much better job at the level and complexity of large codebases. Just my experience.

Comment by pawelduda 3 days ago

Via which software do you use it?