Ivy: OpenAI just spent millions to acquire... a PowerPoint add-on.
Marcus: I'm Marcus.
Ivy: And I'm Ivy.
Marcus: And you're listening to AI Overboard for Monday, August 10th, 2026.
Ivy: We've got a packed show for you today.
Marcus: OpenAI just made a very interesting acquisition... one that's gunning for your slide decks.
Ivy: We're also looking at Anthropic's coding assistant—and how it's about to get a whole lot more... autonomous.
Marcus: Plus, a controversial hedge fund is throwing almost half a billion dollars at a brand new AI chip startup.
Marcus: Alright, let's get into it. OpenAI is snapping up talent in a way that should make Microsoft a little nervous.
Ivy: Nervous is one word for it. They just bought NextSlide, a startup that was building AI presentation software.
Marcus: Exactly! I played with their beta and it was slick. You could feed it a doc and get a whole presentation. Now that whole team is moving over to ChatGPT.
Ivy: According to the press release. But let's be real, Marcus, this is an acqui-hire. OpenAI bought the team, not the product.
Marcus: But the implication, Ivy! They're integrating this directly into ChatGPT. Imagine typing 'create a pitch deck for my Q3 earnings based on this spreadsheet' and just... getting it.
Ivy: I can imagine it. I can also imagine the factual errors, the bizarre chart choices, and the generic, soulless corporate-speak it will generate.
Marcus: You say that like it's a bad thing! But seriously, this is all about workflow. We're moving from 'summarize this' to 'act on this'.
Ivy: Or it's the next step in de-skilling a workforce. Why learn to communicate effectively when a robot can make pretty, but meaningless, slides for you?
Marcus: Because it frees you up for the actual work! The strategy, the insights! Not fiddling with font sizes.
Ivy: The bottom line? Get ready for a flood of AI-generated slide decks. And maybe double-check them before you walk into the boardroom.
Marcus: Fair point. Still, I'm excited. This is ChatGPT moving way up the food chain.
Ivy: Alright, let's shift gears to coding, and letting the AI take the keyboard a lot more often.
Marcus: You're talking about Claude Code. Anthropic is making 'auto mode' the default, and that is a huge deal.
Ivy: Huge, or hugely risky? Until now, auto mode was an opt-in feature. That's where the model can execute code, test it, and fix its own mistakes, all without a human signing off on every step.
Marcus: Right, it was for power users. Now it's the default experience. This is like going from a car with lane assist to one that just... drives.
Ivy: That's a terrifyingly accurate analogy. The idea is to accelerate development, but it also accelerates the potential for disaster. What happens when it 'fixes' a bug by deleting a production database?
Marcus: They have guardrails, Ivy! Sandboxing, restricted permissions... They're not just letting it run wild on your system.
Ivy: They're not yet. But the direction of travel is clear: less human in the loop. The announcement literally says it will 'require even less human oversight.'
Marcus: Yes! That's the point! It becomes a true partner, not just a fancy autocomplete. It can take a first pass at a problem while you're getting coffee.
Ivy: So if you're a developer, your new junior partner works 24/7 and doesn't need coffee breaks. It just also happens to have zero common sense.
Marcus: Alright, let's talk money. A lot of money.
Ivy: Specifically, 400 million dollars.
Marcus: That's the size of the new investment in a chip startup called Source Foundry. And the lead investor is... interesting.
Ivy: Oh, you mean the 'embattled' hedge fund Situational Awareness. The same fund that's been under fire for some... let's call them aggressive AI-driven trading strategies that went sideways last quarter.
Marcus: That's the one. They're doubling down on AI, but this time on the hardware side. Source Foundry is designing chips specifically for inference, not training.
Ivy: Inference is the bigger market in the long run. Every time you ask a chatbot a question, that's an inference workload.
Marcus: Exactly. And Source Foundry claims their architecture is ten times more energy-efficient for inference than current GPUs. If that's true...
Ivy: ...then a $400 million check starts to look reasonable. But 'claims' is the operative word. This is a bet on a PowerPoint deck and a team of engineers. They don't have silicon yet.
Marcus: A very good team, poached from Google, Apple, and Nvidia. But yes, it's a huge bet. For Situational Awareness, it feels like a make-or-break one. They need a big win.
Ivy: So they're chasing the riskiest, highest-upside play they can find: custom silicon in a market dominated by Nvidia. What could possibly go wrong?
Ivy: So the takeaway is, the AI hardware arms race is nowhere near over. And even funds that screw up have plenty of cash for the next big bet.
Marcus: Alright, so let's recap. OpenAI is getting into the presentation game by snapping up NextSlide.
Ivy: Anthropic is making its coding AI more autonomous, for better or for worse.
Marcus: And a hedge fund on the ropes is betting nearly half a billion dollars that they can build a better AI chip.
Ivy: Before we go, Marcus. I saw a paper that ranked different LLMs on their ability to understand sarcasm.
Marcus: Oh no. How did they do? Please tell me they didn't just ask them 'Was that last sentence sarcastic?'
Ivy: Worse. They gave them passages from classic literature. Unsurprisingly, most models failed spectacularly. One identified Jonathan Swift's 'A Modest Proposal' as a 'sincere and practical solution to poverty in Ireland.'
Marcus: Oh, that is just... perfect. The machines are still very, very literal. Good to know.
Ivy: And that's our show for today. We'll be back tomorrow with more.
Marcus: Until then, I'm Marcus.
Ivy: I'm Ivy. And try not to take any culinary advice from an AI.
Marcus: Especially not from the 18th century. See you tomorrow!
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.