2 hours later

new project: there's a bunch of ai slop out there for every use case - img gen, calorie tracking, ai assistants, slide gen, code gen, etc etc. im curating a list of ai tools that actually work with a bunch of friends based on real people feedback. think MTEB leaderboard but for

i got 5 stars on bookface 😭

wtf did i just watch

only in sf

Screenshotting so that it’s never lost

3 MORE DAYS UNTIL THE GREAT LOCK IN OF SEP-DEC

This new DeepMind research shows just how broken vector search is. Turns out some docs in your index are theoretically incapable of being retrieved by vector search, given a certain dimension count of the embedding. Plain old BM25 from 1994 outperforms it on recall. 1/4

first time in life im hesitant to pay for chatgpt subscription (gpt5 is really bad)

tbh the mom test doesn't work anymore - how do you explain to your mom that you're working on "an AI b2b agentic horizontal coding agent thats powered by crypto"

How do you scale infra, enforce strict latency goals, and keep code quality without slowing down? 👀  Had a great time interviewing Findr founder Nishkarsh on the Entelligence AI Podcast! Hear how he handles building a sota vector db as a solo founder!

(can't reply in comments because levelsio has blocked me) ahahah Ethan Lam! nothing before $10mil ARR

Nishkarsh Srivastava is building one of the most innovative solutions in AI But it wasn’t a technical achievement that unlocked growth. It was a GTM shift When Nish first brought Cortex to market, developers didn't understand how it was different from the tools they already

real ones don't let you bcc them until the person they connected you with has responded

Oct 2 - smallest.ai will be sponsoring the VAPICON in SF If you would like to have some cool swag, get insider info, and a lot of fun, come hit us up! Lot of interesting prizes to be won - we have kinda gone over the top for this event. Hopefully we dont get kicked out.

Recently reduced search latency (p50/p90/p95/p100) by over 400ms while increasing indexing throughput 10x with super-optimal memory usage, controlled concurrency, caching, and much more while ensuring compatibility with much smaller CPUs too. At Cortex AI, we're building the

an interface that works while you live Sonatic is out!

Live from VAPICON! Seems like the team really got “Smallest” 😉

this event makes me wish I was in hardtech

bye bye memory tools? docs.claude.com/en/docs/agents…