neil

agents & evals @openai any spice takes are my own

[neil @neil_projects]

quick life update: i’ve joined GTM at @OpenAI!

i’ll be building agentic workflows + evals to help customers turn model capability into production systems.

grateful for the opportunity to build with this team 🚀

[neil @neil_projects]

@nikunj check out 5.5 medium! highly token efficient and 5.6 models will continue on this trend

[neil @neil_projects]

@skirano exactly, models are hill climbing in different domains that are not always reflected by benchmarks. good to have a variety of price/intelligence skews

[neil @neil_projects]

these yc founders are out here connecting codex to their brain because it’s “more token efficient than using words”

[neil @neil_projects]

@trq212 why this over using Cowork?

[neil @neil_projects]

@trq212 how do you know which Claude you’re interacting with? who’s responsible for governing it?

[neil @neil_projects]

@deanwball @OpenAI big fan @deanwball welcome!

[neil @neil_projects]

@NoamShazeer welcome @NoamShazeer!

[neil @neil_projects]

@rileybrown i mean, this is like comparing the cost of a CEO to the cost of a regular employee

CEOs handle the most important tasks and fan out the rest to cheaper employees i.e. subagents

[neil @neil_projects]

my mayor’s jewish
my draymond’s green
my waymo’s broken
i need more tokens

[LakeShowYo @LakeShowYo]

Knicks fans are on a generational run 😭

MY MAYOR STILL MUSLIM
MY BAGEL STILL JEWISH
THE POPE ON OUR SIDE
KNICKS IN FIVE!!! 🖐🏽

[neil @neil_projects]

consumer AI is still incredibly early

1. build something weird
2. delight a niche group of people
3. go viral

[Pankaj @the2ndfloorguy]

i hooked my whoop to my work calendar to find which coworker gives me the most stress 🚨

thanks to fable, I reverse engineered whoop to pull per minute heart rate. nd matched spikes with cal events and attendees

I now have a leaderboard and I think about it daily.

few info masked for obvious reasons ;)

[OpenAI @OpenAI]

We’re bringing new capabilities to GPT-Rosalind, a model series purpose-built for life sciences research at enterprise scale.

It brings GPT-5.5’s agentic coding and tool use together with stronger intelligence for drug discovery, analysis, design, and experimental workflows.

https://t.co/SrAJ3Mt7ka

[neil @neil_projects]

@cjzafir automatic prompt caching can significantly lower the costs. there’s also a realtime mini model that’s cheaper

[Farza 🇵🇰����🇸 @FarzaTV]

Watch me control my computer with just my voice. This is the future of operating systems.

No hands.

GPT-Realtime 2.0 is very, very underrated.

Demo:

[OpenAI Developers @OpenAIDevs]

🧵 Our Voice Hack Night finalists are here.

4 projects. 6 hours. Realtime voice agents in real-world builds.

Now it’s your turn to vote for your favorite. We’ll announce the winner on Monday.

https://t.co/tLllqhl9Tj

[neil @neil_projects]

@mattythetechy1 buy a Fitbit and have codex recreate whoop’s app

[neil @neil_projects]

@MorePerfectUS this was obvious after they acquihired scale ai guy and their top researcher left

facebook’s core product has always been data

[neil @neil_projects]

I can’t stress enough how much of a game changer this has been for me and now it’s completely native in the chatgpt app

[neil @neil_projects]

quick tip: you can spin up a telegram bot that lets you chat with your local codex session from anywhere in about 30 lines of code

[OpenAI @OpenAI]

You've been asking for this one...

Now in preview: Codex in the ChatGPT mobile app.

Start new work, review outputs, steer execution, and approve next steps, all from the ChatGPT mobile app. Codex will keep running on your laptop, Mac mini, or devbox.