We built an internal coding agent. Now 80% of the company uses it
matthieu_bl
13 points
3 comments
November 20, 2026
Related Discussions
Found 5 related stories in 288.8ms across 13,334 title embeddings via pgvector HNSW
- Ask HN: Is anybody producing good code with coding agents? ruffrey · 24 pts · October 02, 2026 · 62% similar
- Six months of writing code exclusively with agents bryanmikaelian · 66 pts · August 27, 2026 · 58% similar
- AI coding has made CI a bottleneck, so we reworked ours to keep up julian_digital · 195 pts · September 21, 2026 · 57% similar
- Harvard researchers find AI coding agents shifted the bottleneck to code review judahmeek · 129 pts · November 17, 2026 · 56% similar
- AI coding agent setups from 663 Hacker News comments titusblair · 19 pts · November 06, 2026 · 55% similar
Discussion Highlights (3 comments)
enether
Pretty good! I’ve been thinking about the multiplayer aspect of AI quite a bit too. An organization inherently needs sessions where multiple people can work on, since that’s literally what every PR or Slack thread is. Integrating a free form, full access AI bot in a PR gets you there like 80-90% of the way imo
basitanwer
One agent with a gateway deciding what each session can reach is a nice idea, almost a decision layer over everything else. What I’d wonder about is the other end: "almost 400 Alaners have created more than 1,700 of them, 1,200 in the last 30 days alone" How do you decide when a prototype has done its job and should be archived? Otherwise it’s space junk that everyone forgets is still up there.
nar001
I always wonder what model/framework someone would use to make all of that, I imagine the results would vary a lot if you chose one model over the other, then you have the concerns about chinese models, etc