Hades

Builder / AI / Robotics / Data

[Hades @hades_blado]

@adcock_brett @grok summarize the video for me

[Hades @hades_blado]

@jeffreyweichsel I think I need to visit your community, lol.

[Hades @hades_blado]

@jeffreyweichsel Oh my, that’s such a great community member! I hope I have a place to raise donkeys and horses.

Haha, we used to have one special community horse, Hercules, when I lived in a ranch. He is a manager of the cattle, haha

[Hades @hades_blado]

As a founder/employee, how many times do you forget your breakfast due to the busy work starting from early morning?

Sad to feel hungry until I realize that I haven't eaten yet

[Hades @hades_blado]

@jeffreyweichsel Cannot bypass simulation if we bet on RSI

[Hades @hades_blado]

@jeffreyweichsel If you are AT&T user, they have a phone upgrade plan, which is great

[Hades @hades_blado]

@poteto For example?

[Hades @hades_blado]

@pelloiafilippo I thought it was because you felt Codex is better, lol

[Hades @hades_blado]

Codex is returning to Cloud. This got me thinking about how agents will evolve. OAI has good models and good harnesses; with Cloud support, it can do many more things than before, and Dots will be a good entry point. Recent popular products like Muse and Instinct brought up a trend of personal agents, but are we really able to clearly separate a personal agent and a "working" agent? A personal agent is a good product strategy, but I feel the boundary will be blurred as the expectation from the users grows. So, maybe the personal agent's scope should be broadened much more. And users will only face one Main Agent (what do you want to call it? my buddy?) for everything.

Some people are using a personal agent to control a coding agent/an agent fleet, which is a good starting point, but this triggers another hard problem--it's very hard for people to exclusively separate personal stuff and work, and I believe no one wants to/has the willingness to spend a long time setting up this stuff because we cannot figure this out ourselves as well. So, from this perspective, the challenge for this Main Agent will be understanding boundaries for different things within the user's whole context, which means memories, personal life, work experiences, context from everywhere, authority checks, permissions, etc. This is like sorting out every line in a messy world. Who can make this happen will possess the market really fast, like how Claude CLI got the dominant position when coding agents became popular.

[Dan Shipper @danshipper]

Codex is one of the most important untold stories in the history of technology.

A year ago, it was a scrappy effort in a quiet corner of @OpenAI. Its cloud product hadn’t found PMF, and the team was starting over from scratch.

Less than 12 months later:

- Codex swallowed the flagship ChatGPT desktop app from the inside (it literally runs as com.openai.codex on your Mac)
- Codex and ChatGPT for Work hit 40M+ weekly users across engineering and knowledge work
- Its leader @thsottiaux was elevated to run all of OpenAI’s core products, including ChatGPT, Codex, and the API

How did they pull it off?
They trained their own dedicated model line, separate from ChatGPT, with researchers and product builders sitting on the same team

They built four different products in a single year (cloud → terminal → IDE sidebar → a desktop app with no code editor at all) as every model leap made a new form factor possible

They used Codex to build Codex—so a handful of people could prove a new product in weeks instead of arguing for it in meetings

I had a front-row seat to the whole arc—from their failed Cloud launch, to OpenAI’s offices the night before GPT-5 dropped, to the day Codex took over ChatGPT.

And I spent months interviewing @gdb, @thsottiaux, @embirico, @mia_glaese, @aambrosino, and @edbayes to get inside how it all went down.

Here is the full, behind-closed-doors story of the team that remade OpenAI:

https://t.co/Ai7lTOJY5r

We’re pushing the boundaries of how stories like this are told.
This piece is completely agent-native—designed to be read alongside your agent inside ChatGPT’s in-app browser:

- Every interview transcript is included in full: Click any quote and you (or your agent) can see the exact conversation it came from.

- Built for your agent to explore: Ask your agent to pull out product lessons for your team, cross-examine the transcripts, or find insights I missed when writing the piece.

It’s a story about the future of software—built to be read with an agent:

https://t.co/Ai7lTOJY5r

[Hades @hades_blado]

@_yuhanluo Thanks for sharing! 🚀 I kept thinking how to integrate human into a synthetic pipeline to boost training. I’ve been on many human projects which generated to much mess and inefficiencies, clearly a barrier for the development under the current speed.

[Hades @hades_blado]

@sama @elonmusk Even tho, you still need to pay the deposit for the car @sama

[Hades @hades_blado]

@alexandr_wang Wanna connect muse to it? 👀

[Elon Musk @elonmusk]

Turns out the future does have flying cars

[Hades @hades_blado]

@poteto @bot Will be interesting to make one for “Who I never reply to on X”

[Hades @hades_blado]

@poteto Lmao

[Hades @hades_blado]

Very wise and elegant lady

[Hades @hades_blado]

Dubito, ergo cogito, ergo sum

[Hades @hades_blado]

Aliens cannot fight the terminator 🤺🤺🤺

[CIX 🦾 @cixliv]

First humanoid deathmatch was absolutely wild