Eli Mernit

Stupidly simple but powerful Claude Cowork workflow. Was looking for office space - normally I'd make a list of spaces that fit some criteria, email the brokers, track the results in a spreadsheet Instead, Cowork takes my criteria and runs this entire workflow for me. A year ago I would have hired someone in the Philippines to do this, now I just use Cowork.

@semil Via Della Scrofa in EV

Q for NYC friends - your personal favorite best-curated small grocery store / deli / bodega in lower Manhattan?

@theovonscousin chess not checkers

@mernit LOL, I just had a long discussion with ChatGPT trying to figure out why they would use OCaml from a technical perspective. It wasn't technical at all.

Jane Street is crazy because their retention strategy is to write everything in OCaml If you leave, you’ll know a lot about OCaml’s garbage collector, emacs, and all the obscure in-house libraries they use But you won’t know any actual company secrets

@faizan10114 @nubank it’s a fascinating phenomenon

jane street was one of the biggest supporters of Ocaml, I remember listening to a bunch of their talks in graduate school. They attracted really smart talent, who would normally not work for a financial trading firm, just because of ocaml @nubank and clojure was the same thing.

guys what if @FisherPrice built @googlecloud

@__morse this is awesome! would love to try it

@mernit this is still very experimental and not ready for production

who is building this?

@whynesspower literally just ask them about their problems and then solve them

@mernit the building part is figured out, how do you actually audit them? you pitch as a consultancy and tell them the ROI of them letting you observe the employees work?

For the past few months, I’ve been running an AI swat team. We learn about a company, audit their workflows, and build them an AI app to automate their busywork. This sounds straightforward, but building these apps is annoying because you keep rebuilding the same things over and over: a filesystem, sandboxes per user, auth, streaming, and connections to messaging apps like Telegram and Slack So today we’re releasing all of it together in a batteries-included Python framework called Capsule It includes a sandboxed computer, multi-channel messaging, background jobs, multi-user auth, and payments – everything you need to ship a production-ready AI product super fast The faster you ship, the more you’ll learn from real customers, and the sooner you get PMF

@tombielecki You can’t automate the process of understanding your customers, but if you read this book enough times the principles become second nature a.co/d/07A3Rpht

@mernit really cool, will try soon. what's your discovery process look like for workflows, do you automate any of that?

@modisulak not if you know what you’re doing

@mernit the "ai app" part is the easy bit, then software shows up and ruins the party

@skyzer4ever you know ball

@mernit Like for medium-rare steak

@DamiDina 🫡

@mernit Nice

@henrytdowling thanks! the only real FDE work is designing the data model and wiring up the Capsule components

@mernit This is really cool! Do you think it's still at the point where FDE effort is necessary to get it up and running for a company? What's the remaining gap?

feel like I'm missing something with many of these AI productivity tools the premise to organize your life, work, projects, etc. is compelling on the surface, but then I sign up and have no clue what to do or where to start

@samuelstros will be curious how it plays out. it's such a compelling idea on the surface.

matches my observations as well. filesystems for agents solve infrastructure issues but are not a (consumer-facing) product on their own