Microsoft to Take On Muse, Trump & Xi Dinner, The Mansion Section | Diet TBPN

TBPN

The episode explores the shifting landscape of AI agents, beginning with Microsoft's transition from Copilot to Autopilot. The hosts then an

Key takeaways

• Microsoft is moving toward 'Autopilot,' an agentic model with its own workspace and memory to compete with Meta's Muse.
• A recent White House dinner included a heavy-hitting roster of tech leaders including Elon Musk, Tim Cook, and Jensen Huang alongside Xi Jinping.

Main topics

• Microsoft Autopilot vs Meta Muse
• White House Tech Dinner Diplomacy

Notable quotes

Microsoft is about to have their Muse moment.

Conclusion

A comprehensive look at how AI, geopolitics, and economic shifts are f

Transcript preview

Speaker 1 (0:01) We got some debates on the timeline. I identified four debates. You don't want to call them debates, though. You want to call them arguments, fights? I thought Speaker 3 (0:10) arguments. Okay, Speaker 1 (0:11) we're throwing down. We're throwing down. Speaker 3 (0:13) Wouldn't someone rather listen to an argument than a debate? First Speaker 1 (0:17) one, Microsoft. Is Microsoft about to have their Muse moment? They've launched Copilot. Now they're launching a new product today. Is it going to be a breakout success? Is it going to compete directly? with Meta and Muse, or will it be a different thing entirely because it lives in the enterprise, it lives in the Microsoft ecosystem? Let's go to Tay Kim. He says, shots fired. Microsoft will go after Meta's Muse. From Alex Heath's interview of Satya Nadella, he's doing the rounds. Quote from Alex Heath. He wants to bring Autopilot's AI chief of staff to your personal life too. It's not staying in Excel. It's not staying in Outlook. It's going to be. Cooking dinner. Big Speaker 3 (0:57) move going from a co-pilot to autopilot. Autopilot. Yeah. It says a lot about their intentions with the product. Speaker 1 (1:03) Yeah. Yeah. It is actually sort of the, like, is this AGI? Is this not AGI? Like, if you're in the co-pilot mode, you're sort of signaling we're not really in the AGI era. Autopilot feels more like, although it's a term, it's a vague definition, it is significant in the brand term. So, it's a hardened version of OpenClaw. The agent gets it. gets its own computer workspace and memory to work continuously. Virtual machine for everybody. It'll have its own computer workspace and memory to work continuously. When I ask about Microsoft's consumer strategy, he points to its 100 million plus consumer subscribers. Microsoft broadly, they're going all in with autopilot. Autopilot should also go to the consumer side. He sees consumer agents potentially cutting out today's middlemen. making that market more zero-sum. In the enterprise, he thinks agents will make the market bigger than cloud by orders of magnitude. We got room to run. Satya said it himself. He's going to 300T, $300 trillion company. I like it. What do you think? Speaker 3 (2:12) So first thought is that, again, everyone is building the exact same thing. So we went through this for the last few years. Everyone was billing Speaker 1 (2:20) it. Speaker 3 (2:20) Everyone had a chat Speaker 1 (2:20) app. Speaker 3 (2:20) Everyone was billing it. The funny thing is Speaker 1 (2:22) the lore on GPT-4, that was first available in Bing. GPT-4, the thing that, like GPT-3. GPT-3.5 powered the original chat GPT. GPT-4 was only available through Bing. And then everyone was like, oh, I guess Microsoft's going to make a play and Bing's going to be doing AI and this is going to be a challenge to Google. And then they sort of pulled back. Well, Speaker 3 (2:45) and at that time when Microsoft was backing up the Brinks truck for OpenAI, they certainly... I don't think it was, it certainly was non-obvious at that time that they would be creating a company that they would ultimately compete with as much as they are partners. Peter Speaker 1 (3:01) Steinberger, the founder of OpenClaw, chimed in. He said, Microsoft shipped a really compelling product on top of OpenClaw today. We worked with them since March to make the code base ready for large-scale deployments. And of course... There might be some IP sharing agreement, right? Because didn't, oh no, because OpenAI hired Peter, but they didn't acquire OpenClaw. Yeah, OpenClaw is a... Open source. Yeah, non Speaker 2 (3:21) -profit. So Speaker 1 (3:21) anyone can build on top of it. Yeah. But what's odd is that I think Nat Friedman stated that they didn't fork OpenClaw. They built something from scratch. But... There are just conventions. And so you wind up having similar looking .md files. It's just like if you build an e-commerce site, you're going to have a product detail page. You might have a similar site map and architecture. It doesn't mean that you forked Shopify to do that or whatever. So the debate point is how big will this be? What is the actual impact to Microsoft? There is the cool. macro tech story, which is just that the models have advanced to a point where there is a new use case. Like they're useful in a new way, not just knowledge retrieval, not just go and write a bunch of code. They are useful. But that requires a new type of harness. Just like when the models got good at code, the harness became very important. We saw Cloud Code and Codex and Cursor and a bunch of other companies, Cognition, really define that era. Now the models are pretty good with personal agent stuff, sending text messages, talking to people, integrating with Slack, et cetera. So the harness and the integrations and the... partnerships matter. So we're talking about partners with Amazon. Who's going to be Amazon's dance partner? Who's going to be Walmart's dance partner? Who's going to partner with Shopify? How big is the open ecosystem? You need a different harness. You need different integrations. And you need a lot of new user education because I think there's a lot of people. I mean, there's people that are still online being like, these things hallucinate. They have a knowledge cut off. Yeah, the Speaker 3 (4:55) main thing with this launch is that I don't think there's anyone on X. that seeing Microsoft launch an open-claw powered agent and thinking, I've got to use that, I'm so excited, I really want it. That's my view, and I think it's correct, and you're free to steelman it. Speaker 2 (5:11) That Speaker 3 (5:11) being said, they have such incredible distribution, they can still roll this out to a bunch of people who are going to try it and be like, wow, that's really cool and useful and helpful, and be the first way that they experience this new paradigm of agents. And so as much as I just think there's... basically zero excitement from like the core insider agent maximalists. There's going to be just natural excitement from everyday windows and Microsoft users. Speaker 1 (5:37) Yeah, it is uncommon in tech, and maybe it's uncommon in X to run your entire organization, your enterprise on Microsoft. The default tech stack for most startups is Google Enterprise, Gmail. You set up the Gmail account for your employees, and then you add Slack, so then you're in the Salesforce ecosystem. And Microsoft is there, but usually with one-off Excel licenses here and there for the finance team. Speaker 1 (6:07) resonate with you in terms of like how these products like fit together there's usually some microsoft but you have to go to a financial company or like a real economy company to see like okay these guys are all in on microsoft as a tech stack like outlook down to bang down to the databases and fully integrated into Microsoft. It feels like it could be more successful than the old copilot, and a lot of that's driven just by capabilities. The models are better, the harnesses are better, the integrations are better. And so the big question is, when is Google going to fire back? Because are they going to wait until the next I.O.? That's a long way away, but they do love releasing on. you know, those annual schedules. So it's like, we might get a Google, like a crazy Google open claw agent, personal agent in mid 2027. And then Apple's close behind in like 2035, something like that. They would launch something. I mean, Google does have, there is Gemini Speaker 2 (7:05) Spark, which is like a 24 seven always on agent, but like no one. Does Speaker 1 (7:09) it have a VM? Does it have a real virtual machine? Speaker 2 (7:13) I mean it says even if your phone and laptops are turned off it'll still work. But I've seen like very little, like no one's really talking about this. Speaker 1 (7:20) Anyway, it'll be interesting to see if there's like, you know, a niche community of like, yeah, I'm a Microsoft autopilot guy. I mean the COD integration could change things. It just depends on how you're positioning it. Like if the brand, if you talk