not a human in sight

agent-driven software is going multiplayer and asynchronous websockets are set to dominate 2025

i believe this is the future of software: personal, malleable, friendly software should adapt to people, not the other way around

99% of apps need to implement these in the same way ideally, these can be handled in advance and abstracted so the model’s finite complexity can be spent on your biz logic, not common infra shipping something based on this premise soon

incredible talent density in the office today we need to ensure these founders have everything they need to stay and build right here in Chicago today, tomorrow, and for the next decade massive props to Landon, Hank Couture, Danny Goodman, and others working hard on this

it shouldn’t be this hard we have much smarter models than we have context collectors we have the terminal, we need the mouse

Bighorn Sheep are not a Bioweapon.

late to this, but nobody seems to have said it: all of these logos are systems of record that benefit from agent interactivity with their existing systems very few agent companies here for a protocol for agents to talk to agents curious

i can't tell if this company (e2e advertising agent) is good or bad at their job they're getting clowned all over my feed but they're all over my feed

as the cost of software development goes to zero, distribution becomes the bottleneck those with existing niche distribution are poised to win as software becomes content

mode fondateur

early 20s male startup founders when the frontal lobe finally kicks in

Prism (Prism) is an AI that watches humans use software and tells developers what to fix. We have the smallest recording package on the market and require zero instrumentation to set up. ycombinator.com/launches/NVC-p… Congrats on the launch, Rajit Khanna, Land Tantichot, and

Chroma and Turbopuffer going to war

funny prompting hack–including just the phrase: chekhov's gun keeps models focused and logically consistent through a longer response (caveat: responses will include mentions of "gun" and "bullet")

bruh

if you're sufficiently ambitious building something new will go something like this: V1: for exploring V2: for learning something important and unintuitive ~V3: breakout don't try to rush this process. going through it creates understanding which forms the beginning of the only