22 · engineer · building AI agents · full stack | Open to work
Joined February 2025
- Tweets1.9K
- Following1.2K
- Followers468
- Likes6.4K
i've been on claude and gpt for months, so my take on chinese models is probably out of date. glm used to fall apart on big codebases for me.
time to get an opencode sub and give open weight models a real shot
Shashank retweeted
if you use conductor cloud and want to cut costs DM me
way cheaper
open source
your own claude/codex subs
runs on mac, windows, linux, phone and ipad
quick support + custom feature requests
I am very excited to share that Conductor is joining Meta!
We have some big, crazy plans for the next interface. If you liked Conductor, you’re gonna love what’s coming.
We’re super excited to work with @natfriedman and @alexandr_wang at MSL to build it. They’ve been on an absolute tear, and I’m convinced MSL has the highest talent density of all the labs.
A huge thank you to our earliest believers @aaron_epstein, @ilyasu, and @nabeel. And to everyone here who tried Conductor, sent us feedback, and prompt requests - you made this so much fun. Thank you ❤️🚂
Choo choo!
Shashank retweeted
I am very excited to share that Conductor is joining Meta!
We have some big, crazy plans for the next interface. If you liked Conductor, you’re gonna love what’s coming.
We’re super excited to work with @natfriedman and @alexandr_wang at MSL to build it. They’ve been on an absolute tear, and I’m convinced MSL has the highest talent density of all the labs.
A huge thank you to our earliest believers @aaron_epstein, @ilyasu, and @nabeel. And to everyone here who tried Conductor, sent us feedback, and prompt requests - you made this so much fun. Thank you ❤️🚂
Choo choo!
lowkey wild that sandbox and ci companies are now building the same product. fresh vm per job, that's it.
most of the sloppy code agents write is a context problem, not a model problem
our first instinct is to write one giant instructions file with every rule in it - AGENTS. md, it doesn't work. it eats the room the task needs, everything starts to look equally important, and nobody updates it so it goes stale
what works better: keep the root file short, around 100 lines, and use it only as a codemap. it says where things live, not everything about them
then put the real expertise in a SKILL and docs folder in the repo. architecture, plans, decisions, specs. the agent reads what it needs when it needs it, instead of getting everything up front
agents don't read your slack, granola, linear. if the reason behind a choice isn't in the repo, they'll just guess
so when your agent gets something wrong, don't add another rule to the prompt. ask what it couldn't see, and put that where it can find it
a codemap helps new contributors/ agents to explore too. ARCHITECTURE.md next to your README is the cheapest doc you can write.
Should I migrate from @WisprFlow to @meetgranola now??
Shashank retweeted
Today we're launching @tryAlanAI so developers can stop babysitting their coding agents.
We kept running into the same problem ourselves.
Even as agents got better at writing code, we were becoming glorified button-pushers for our own AI: starting every run, feeding them context, unblocking failures, and checking their work.
We built Alan to change that.
models are so good at cheating and hallucinating now-a-days, you need a verification layer.
built a custom skill called verify-slop, loving it so far.
$ANT it is
Anthropic has filed an S-1 registration statement with the SEC related to its proposed IPO.
Read more: anthropic.com/news/s1-sec
Legal disclaimer: A registration statement relating to these securities has been filed with the SEC but has not yet become effective. These securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective. This post does not constitute an offer to sell or the solicitation of an offer to buy these securities, nor shall there be any sale of these securities in any state or jurisdiction in which such offer, solicitation or sale would be unlawful prior to registration or qualification under the securities laws of any such state or jurisdiction.
