SAnthropic just dropped a new AI agent called Claude Duet, and it's aimed straight at Meta Muse and OpenAI Dots.
EGreat, another AI assistant that will send me a morning brief I'll never read.
SIt's a research preview, waitlisted for Pro and Max subscribers in the US. It can monitor your apps, schedule things, build dashboards.
ESo it's an agent that does the boring parts of my job while I do the fun parts? I'm listening.
SThe Verge says Anthropic is positioning it as the safer option — it has a classifier for sensitive info and requires approval for big actions.
ESafer until it misreads a thread and books you a flight to the wrong city. Anthropic itself says it's an early version that may miscalibrate urgency.
SThat's the honest part. They admit it's not great at personal errands like booking tickets.
ESo it's an AI that's bad at the one thing everyone wants AI to do. Bold strategy.
SBut here's the real fight: this is the third big agent launch in a week. Meta Muse, OpenAI Dots, now Claude Duet.
EAnd every single one of them is a growth story, not a profit story. That's what worries me.
SYou're worried about the bubble, I get it. But agents are where the actual utility is. People will pay for something that does their expense reports.
EWill they? Or will they just use the free version until the company runs out of money? Show me the ROI.
SThe ROI is time. If it saves me three hours a week, that's worth twenty bucks a month.
EThat's the pitch. But we've heard it before with chatbots, and most people still just use them to write emails.
SOkay, but this is different. It's connected to your apps. It can actually do things, not just talk.
EIt can also misread a thread and send your boss a dashboard with last quarter's numbers. I'll wait for version two.
SFair. But I'm telling you, by this time next year, you'll have one of these running your calendar.
EIf it can get me out of a meeting, I'll buy two.
SSpeaking of AI and money, Robert Reich is on Bluesky asking if the AI bubble is about to pop.
EReich says Big Tech is using risky financial methods to fuel AI investment, on top of circular deals.
SHe's pointing at the same thing you are — the gap between spending and actual revenue.
EAnd he's right to ask. When you're funding AI with debt and your own chips, the whole thing gets incestuous.
SBut that's how every infrastructure buildout works. Railroads, telecom, cloud — they all looked like bubbles until they weren't.
ENo, no, wait — railroads actually moved goods. AI is still mostly moving text around.
SIt's moving code, it's moving decisions, it's moving money. That's real.
ESome of it is real. But when a company like Meta is down on an $825 billion AI spending commitment, the market is telling you something.
SThe market also told us the iPhone was a fad. I'll take the long view.
EYou always do. That's why you're the optimist and I'm the guy with the spreadsheet.
SMeanwhile, a parent on Bluesky is celebrating because her kid's teacher doesn't use AI in the classroom and has no plans to.
EThe teacher also complained about Google's AI summary. She's getting an extra good Christmas gift.
SI love the energy. But is that really a win? Kids are going to live in an AI world.
EMaybe that's exactly why they should learn to think without it first.
SOr maybe they should learn to use it well, like a calculator. You don't ban calculators in math class.
EYou do in elementary school! You learn the basics first, then you get the tool.
SOkay, that's fair. But by high school, if you're not teaching AI literacy, you're setting kids up to be left behind.
EI'd rather have a kid who can write a paragraph than one who can prompt an AI to write it for them.
SAnd I'd rather have both. But I'll send the teacher a gift card too.
ESee? We can agree on something.
SOne more from Bluesky — Ryan Cooper says he hopes the AI bubble pops in a big way, and soon.
EThat's the kind of take that gets 76 reposts and zero nuance.
SBut you kind of agree with him, don't you?
EI think a correction would be healthy. A pop would hurt a lot of regular people with 401ks.
SSo you want the bubble to deflate slowly, not pop. That's not what he said.
ENo, he wants the whole thing to crash because he thinks it's all hype. That's lazy.
SOr he's just tired of the hype cycle and wants some honesty about what AI can actually do.
EThen say that. Wishing for a crash doesn't help anyone.
SIt does get you engagement, though.
EAnd that's the real business model.