Apple’s Siri AI Comeback Test?

**Karl Yeh** (0:00)
Hey, what's going on, everybody? It's Monday, June 8th, 2026, and this is The Daily AI Show Live. It's a new week. It's episode 741 of The Daily AI Show, which is always nice to think about. And with me so far today is Andy and Brian. I think we're gonna see Garris maybe popping the door here in a little bit as well to help support.
So it's, you know, Monday, Andy, and I wasn't here on Thursday or Friday, so I always feel like, you know, perhaps I missed some of the news on there. Morning to Cisco and Japanet in the comments as well. Appreciate you guys being here. But I know one of the stories that seems to be a bit of a, recircle, I don't know what the word I'm looking for here is, but it's come back around again, is this idea of OpenAI, Sam. I think it probably, if I had a guess, you may know better than me. Maybe he was just in an interview and talked about this again. Sometimes that's what hits the news cycle. But regardless, one of the things I saw was, hey, ChachiBT, if they keep that name, I wouldn't be shocked if maybe they didn't or whatever, but they're trying to consolidate that Codex is going to come back in underneath it, that Atlas is going to come back underneath it, and it's going to be this one-stop shop, which to me again, feels mostly like what Google does very well, actually, is that they have an entire ecosystem, and especially we just saw with 3.5 Flash. Now, lots of people have opinions about how good 3.5 Flash.
Ethan Malik was talking about how he feels like for the Pro models, which are still at 3.1 Pro, does feel like it's lacking, whereas only four or five months ago, it was actually a premier model. But because things are moving so fast these days, that it does feel a little behind. As much as I like Gemini, I would agree with that, I would agree also that Flash is not a great default model because I think it's missing some of that premier level, albeit it is very quick. It oftentimes tries to hoover up everything at once and just throw it out there. Anyway, the story is more about like what I guess Sam had talked about how they're going to bring anything together, where Greg Brockman fits into it, how they're going to organize it. I'm just curious. I wanted to bring it up in case you hadn't, but did you guys talk about that on Thursday and Friday as a new story?
Why this is re-circulating again? Was it from an interview or anything?

**Andy Halliday** (2:27)
Well, I think there are a lot of analysts who are trying to parse what the product strategy is for each of the main players who are competing with each other to get the largest and most persistent catchment of users out there. Now, OpenAI has achieved a lot by being the go-to chatbot interface. So people using chatbot-like features to answer questions and to provide sort of minimal level search, that's one mode and they've certainly dominated that at least with their free usage around the world.
To some degree, I think that that chatbot function has been somewhat separated in the early stages of this from the agentic models that are out there. So you're not going to the chat interface to get any kind of agentic action. You have to go to Codex or Claude Code or something like that. And Claude Code and Codex were doing agentic work in coding, which was the principle and first application space that these things were being used in, more advanced models with advanced reasoning and ongoing planning, and then execute and then evaluate kind of modes.
But then both Claude and OpenAI with Codex said, wait a second, these agents can do much more than just coding, so why don't you use Codex for everything? Right. And similarly, Claude Code said, well, in our application, let's put what we've learned from Codex behind the scenes and let's do some code work and give an agent access to your machine and let the Claude powerful reasoning model manage all of that.
Now, OpenAI, I think, has telegraphed that their ongoing strategy will be to try to unify all that and come back to the idea that you'll just go to one place to get anything done. And that's different from Google strategy. Google strategy is you're already going here and there and there in our Google ecosystem. We're just going to power those by AI now. And each of those AI empowerments might be somewhat different from the others. Whereas OpenAI and I think ultimately Claude will have their primary powerful foundation model as the conversant with you and distilled versions of it for a shift issue and so on. But there's that other model which is I just I don't want to have to manage all these other different functionalities and application use cases. I just want to talk to my assistant and I want that AI assistant to do everything for me. That's why I think it's coming up now is that's where we're converging.