ChatGPT Atlas

Anyone feel like OpenAI is acting like Google lately? They announce a lot of products/features and then kill them when they realize people don't use them[0]. They also announce products way before they're ready for launch, just like google[1].

- GPT Plugins? (HN went crazy over this, they called it the "app store moment"...)

- GPTs?!

- Schedules?

- Operator?

- The original "codex" model?

[0]: I know, the diff is that google kills them despite knowing that many people use them.

[1]: I know, the diff is that google sometimes doesn't launch the announced product at all...

I think it's more like Apple - use ChatGPT as the iPhone and build an ecosystem around it

How else would you build new products?

I've been using Schedules a little bit, but yeah fair.

This is exactly the right move to find out what products people find useful on top of their AI infrastructure.

They don't care. They KNOW they want to be Google.

Even if it means throwing away their experiements. That is how you test if a new product works or not.

The difference is, they have over $40B+ in funding, meaning they they can afford to do that.

At least Google waits a few years before killing things... OpenAI barely lets the paint dry :)

building and shipping are not the same.

You're supposed to invest 10% ~ 20% of your resources into life support for dead-end products for eternity

I don't think they've done much that's that impressive since the launch. Cancelled my sub ages ago for other offerings.

Zero moat with crazy amounts of debt and financial engineering. What could possibly go wrong?

yeah the custom GPT announcement was literally a carbon copy of steve jobs announcing the app store down to the mannerisms and tics. You could tell they were doing it to give VC and private equity the pavlov bell "ring ring! this company is apple!" and remember the sora announcement just a couple weeks ago? oh you guys are tiktok and instagram reels now too? cool...

everything that openAI does is laser focused on valuation valuation valuation

of course it's a weird form of valuation because like remember when these guys are a non profit? lol

It is weird though, the "I'm a bigtechco dance" seems to be working, even though the economics on providing LLM services do not in any way justify the valuation.

they have like five five credible competitors who are right behind them BTW

building, shipping, and then discarding when it becomes apparent that the product doesn't have a future, is about as good as it gets. most businesses have trouble doing just one, let alone all three.

how is codex on this list though? "agent" is also alive for now, and I doubt it's gonna go down any time soon

They're doing a lot of cool experiments and don't mind discarding them when they feel like LLMs are moving in a different direction. I don't know why people are complaining here. Every time I read AI posts here there's like an army of commentators that seem to have little AI usage.

They have too much money and need to do something with it.

A lot of this is about building an ecosystem. Just a good LLM won't be enough forever.

But if you have a giant network of products that *only work with your other products, you might become the next Salesforce.

> They announce a lot of products/features and then kill them when they realize people don't use them. They also announce products way before they're ready for launch

Agreed. They really should have named this product Atlas *shrug* ¯\_(ツ)_/¯

They haven't actually canceled most of these things.

I think its funny how its a Chromium-based browser, and the live demo involved automating Gmail and Google Sheets. Google literally runs the world; we're just playing on their playground.

Maybe it depends on a person, but I find some of their products quite useful. For example, I use a few of my own custom GPTs almost daily and have a few scheduled tasks running.

What's with the assumption that everything needs to be a "moat"? Seems much more important/interesting to wire up society with cohesive tooling according to Metcalfe's law, rather than building stuff designed to separate and segment knowledge.

It's not that weird. Both of those examples are probably some of the most common automations people will be using.

Gmail and Google Sheets has not cannibalized the existence of alternate email providers or spreadsheet programs, we can relax a bit on those fronts. AdSense, on the other hand...

General advice on Hacker News, social media for YCombinator startup incubator is as following:

1. Start New Company

2. Hire first employee: security and compliance engineer

3. Finish security audit

4. Post security bounty program (10% of gross revenue for finding security@company.com email)

5. Use only real languages like C, but intrinsic are bad. If you want to use intrinsic, use x86_64 assembly language.

6. Any time anyone suggests hiring sales guy, hire another security engineer, increase security bounty 10%

7. Start on initial MVP. Pre-commit hook send every patch to security engineer. Once he has reviewed you may commit on used Thinkpad.

8. After twenty years of this, bootstrapped, you have Hello World triangle display on screen. Congratulations.

9. Publish 100 Year Support Program: anyone who buy Hello World program entitled to full discount within 100 year of purchase and given source code

10. ???

11. Profit

Very smart advice from entrepreneurs of Hacker News

I remember custom GPTs also being touted as the "app store moment" for ChatGPT. OpenAI even had big plans to pay creators of custom GPTs, but it seems like that never really materialized. I think people quickly realized that custom GPTs are more or less useless, and certainly not something that would ever drive revenue.

> yeah the custom GPT announcement was literally a carbon copy of steve jobs announcing the app store down to the mannerisms and tics.

And then they nabbed Jony Ive of all people for their hardware project, with Altman stating that Steve Jobs would be "damn proud" of what they're working on. It's about as subtle as a brick to the face.

Dislike of the hype does not mean people don't find value in the tools.

> everything that openAI does is laser focused on valuation valuation valuation

idk it seems like a company filled with product and engineering where people are thinking of cool product ideas and shipping them. They don’t have to all hit, but it doesn’t seem bad to try them.

This is what happens when one third of your employees is Xooglers

All these “product ideas” are an irrelevant waste of time if they actually develop AGI.

To me it comes across as them hedging their bets that the snake oil Sam Altman has been selling might not actually pan out.

Agree these are the most common. And bad for OpenAI is that both of these have Gemini heavily woven into them.

the company charter is not product and engineering it's "we are going to invent THE machine, the singular fulcrum upon which the infinite lever of history, the transcendent union of man and machine, the birth of a new GOD, rests."

I mean, ok, you're product and engineering, fine. You get $20/mo out of your million or so paying users and $200/mo from a small handful of freaks. what does that mean for the valuation? what does that do for sama's patek phillipe collection?? nothing good I assure you. the AI product and engineering landscape is insanely competitive, like actually competitive.

that's what I'm saying, the circle doesn't square here.

btw the product ideas ARE cool. I like them. they are not worth eleventy trillion dollars.

Plugins is the only thing that flopped though. What do you mean by "GPTs?!". Schedules are useful, Operator could be the next version of Atlas.

"OpenAI has zero moat in LLMs"

Yeah, it's the hottest tech company around right now, so there's probably a lot of parallels to Google in its heyday.

We're in a bubble, a mania, and the top company is excited to announce they're gonna do erotica.

It is hemorrhaging money, with little in its offering to distinguish its competitors.