Enjoying this issue?
Get tomorrow's AI & engineering digest in your inbox — hand-picked, summarized, and always spam-free.
TLDR
ChatGPT Spaces, a new feature announced at OpenAI DevDay, aims to compete with Google Docs by embedding AI agents directly into documents, making collaboration more seamless than traditional tools. This was the most developer-relevant release, as other announcements leaned heavily toward consumer products and enterprise features without providing crucial technical benchmarks.
Key points
OpenAI did not provide benchmarks for the Realtime API compared to competitors.
The release of a new GPT model variant came only days after the initial launch, an unusual cadence.
Two live demos failed during the keynote, using pre-prepared examples instead of real interactions.
Codex Cloud was announced but only supports GPT models and is not open source.
Tools mentioned
Stop scrolling. Start reading smarter.
Receive the day's most important AI & engineering updates in one concise email. No spam.
Transcript (captions)
All right, so we're at the Open AI dev day. Yeah. >> What do you think about the announcements, David? >> So, it's kind of mixed, you know,
doesn't feel like it's for developers. First of all, a lot of this is just consumer products or like enterprise features and just general like products and releases. I wish there was more like
technical stuff, you know, when they released their decision API, which is a competitor to Jeff. Where is the benchmarks? They should have shown how it compares to Jeff, right? If you if
they don't show the benchmarks, it makes me think like might be worse or way more expensive, right? It also is weird with the timing of GBD Soul 6.1 literally like two or three days after GBD6 soul
which like that's very strange that never happens basically in AI. So it makes me think like they did that because of anthropic but then none of the benchmarks they show how it compares
to Opus 5.5 or Sonet 5.5 which released yesterday. So very like very strange so far. It doesn't really feel for developers just feels like they're letting you know kind of this corporate
creep go in there. also two failed demos which again obviously it happens live demos but like given the fact that they were using like pre-prepared fake examples and not like real stuff. I
don't know. I don't want to be only negative, you know, we've met some great people, but it doesn't feel like anything's too much of a breakthrough for us. I think the main thing is
honestly the spaces. So CHBD space is kind of like a competitor to Google Docs which you know again Google Docs is easy for humans but adding an agent there is inconvenient and you know if you can
have Astra or Soul or whatever interact in a file a document they're also going to do presentations it's definitely going to be the the way to do it. So yeah I think that's probably the main
release for us and also they released the cloud which I codex had cloud what exactly did they release with cloud? Not exactly sure to be correct. >> Something was cloud. Codex already had a
cloud. Um, so I don't know what exactly they released. But one thing is for sure, it's not open source and it will not allow you to use any model. You're only going to be using GPD models in
there. So when it comes to loud room, I literally have zero feelings and like I don't think it impacts us at all at all. A lot of these companies when they get too big, they want to keep everything
closed. They want to keep everything like ecosystem lock into their own subscriptions, their own inference. And that creates tremendous opportunity for startups. And I think this is just a
cycle that will keep repeating forever.