nsoybean

5. AI Meeting Memory: Desktop app to transcribe meeting calls github.com/nsoybean/ai-me…

4. Switchboard: Parallel coding agent workspace with native terminals and git worktrees github.com/nsoybean/switc…

3. Ship Social: Convert GitHub releases, PRs, and commits into social content on X github.com/nsoybean/ship-…

2. Solostack: Local-first runtime for managing an AI company with agents, orchestration, git workflows, and CLI-driven execution. Uses your own claude/codex subscription. github.com/nsoybean/solos…

1. Nira AI: Personal assistant. Use it via webapp or telegram github.com/nsoybean/nira-…

Open sourcing all my AI projects that i have done over the past 1.5 years

First time in an autonomous car in Guangzhou, China 🇨🇳 Going at 40mph and frequent lane changing is not something I’m expecting 😂

One week break from coding 😮‍💨

@VivekNairbx What solutions have you tried?

@nshawbin managing context

For those building agents, what’s a challenge that you can’t solve with existing solution? I’m happy to try and build them 👷‍♂️

Oh wow

Nicely written on how to improve your loop in practice

@agrimsingh @llama_index wow, maybe pdf can be git versioned now

i built the fastest pdf parser available (and it's OSS!) glyphrush: 400+ pages/sec, and every page it can't fully handle comes back flagged - requires_ocr, layout_uncertain, table_uncertain - instead of silently broken text. benchmarked against @llama_index's liteparse aka the "world's fastest PDF parser" on its own benchmark design, same machine, warm and cold. glyphrush is faster in every cell: 2.26× vs its no-OCR mode, 110× vs its defaults - while passing the same 9/9 quality gates on 8,624 labeled checks. built with @cursor_ai @ericzakariasson @leerob

@zachlloydtweets do you think this replaces workflows with deterministic scripts as verification gates? since now its purely based on verifier agent evaluating the claims of coding agent

If you use /loop or /goal for long-running agent sessions: how do you verify the agent actually did the thing, vs. narrated doing it? /goal's evaluator uses a small model to judge the conversation, not the diff, nor run deterministic scripts Curious what people's setups are — hooks? CI? manual review?

Building a running route planner and hit a wall 🧱 @OneMap_sg only supports walking/cycling/car modes - so the routes cut straight through HDB blocks. Not exactly the vibe for a morning run. Any SG devs found a workaround for runner-friendly routing? 👟

@brianchew I thought my keyboard spoilt

i remember the first time i came across vim and 🤮🤮🤮🤮🤮 :wq! !!!!!!!!!! (sorry vim people)

I like this answer. Basically, in my own understanding is trying to systemize your process. Up until recently people have been defining their workflow for building feature X. I would think the idea now is to move one level up, and come up with higher level workflow to handle X, without knowing in advance what is X. Eg. If it’s a bug, it will be reproduce, write test, build, run test, review etc. If it’s feature request, then another flow etc

Non-technical idiot guy here. What does this mean for the non-coder audience? I truly want to feel what Boris and Peter are talking about - anyone got any resources to help bridge me there? Thank you!

@hridoyreh How did you get these revenue numbers?

Indie hackers run boring sites. They are making $1.7M+ per month. 12 boring websites: