When Claude for Legal launched, it briefly moved the stocks of Thomson Reuters, RELX, and Wolters Kluwer. That's a harder test than any benchmark we've been arguing about all week. This is the Tech Podcast Podcast — and today we've got a Kleiner partner on Bloomberg, a legal AI promo, and a founder who rebuilt the same company three times on fifteen lakh rupees. Barry Ritholtz over at Bloomberg has the details. So the pitch here is Mamoon Hamid on Bloomberg — nine years in early-stage AI at Kleiner, early in Slack, Figma, and Glean. Ritholtz is hosting, and he actually presses people, so I'm curious if Hamid gets pulled off the greatest-hits reel. Exactly. It's a macro-finance audience, not a founder audience — so what AI framework does he offer that the 20VC or All-In crowd hasn't already run into the ground? And here's my ask: can Ritholtz get him to name a loser? Anybody can list Figma and Glean after the fact. I want the layer Hamid thinks gets squeezed when the model commoditizes. Right, because Kleiner sits in the same AI dev-tools lane as Cursor and Nebius, the names Accel was taking credit for yesterday. Coming from a rival, naming the squeeze would actually mean something. A brand-story victory lap wouldn't. From Rohin Dharmakumar at The Ken:
Everyone thinks Shiprocket began in 2017. But the company Saahil Goel and his co-founder Gautam Kapoor started was born in 2011, and it was rebuilt three times before it worked: a “Shopify for India” called KartRocket, a marketplace called Craftly, and finally the shipping-and-enablement layer now behind a quarter of a million Indian sellers, on its way to a public listing.
Okay, this is the one I actually wanted today. Fifteen lakh rupees, and the company got rebuilt three times before it worked — KartRocket, then a marketplace called Craftly, then finally the shipping layer. A quarter of a million sellers now, and it's heading toward a public listing. The framing is that everyone thinks it started in 2017, when it actually started in 2011. Six years of failing at the same idea in different costumes before the third rebuild stuck. That's the operating detail I want — not the IPO deck version where it was always destiny. What grabs me is the reason Indian SMBs wouldn't pay for software. That's a concrete constraint that killed the Shopify-for-India version — and it explains why he ended up under the transactions instead of selling seats on top of them. Right, but that 'one paisa of every Indian transaction outside the marketplaces' line is a hell of a slide. My question for a two-hour First Principles sit-down is whether Rohin gets him to say what the actual moat is, or whether it's just a big number said confidently. And there's an investor ultimatum in there — take the four million dollars or nothing. After a Kleiner partner spent our last segment talking term-sheet philosophy, it's grounding to hear it from the founder side of that standoff. This one's from Gainsight:
The founding CRO of HubSpot, Harvard Business School lecturer, and Stage 2 Capital co-founder joins Josh to break down his new book, The Science of Scaling. The argument at the center of it – the decision of when and how fast to scale shouldn’t be a gut call. It should be gated on retention.
Mark Roberge — HubSpot's founding CRO — says treating product-market fit like a feeling is as absurd as calling profit a feeling. Bold, coming from an industry that runs entirely on vibes. But he actually gates it on retention, and he names a leading indicator — one metric that predicts churn in a customer's first month. That gives you something you can test, instead of just a gut call. The detail I want is Drift's founder flying across the country to onboard fifty-dollar-a-month customers. Forget growth multiples for a second — that's a guy learning why people churn by hand. And if retention is the signal, it reframes those VC growth benchmarks — the 4x-to-5x floor everyone's been quoting all week is measuring output, while Roberge says the gate is whether anyone stays. Different question entirely. Right, growth multiples are the scoreboard. Retention's the thing that tells you if you should even be on the field yet. Tushar Jain over at Bankless has the details. So Multicoin's title is 'backend for ALL of finance,' and the very first thing you hear is David reading a disclosure that Multicoin owns hype tokens and stands to gain if the price goes up. Just want everyone to hold that thought while we discuss the moon. The claim under the hype is more interesting than the title. Perps are just the wedge. Get the exchange, then become the settlement layer everything else runs on. Which has the same shape as Alexandr Wang's 'personal agents run everything' pitch — one thing becomes the backend for a whole category. Except this one at least has an architectural argument for why perps get you there, instead of just a vibe. It's the pipe question again. One chain owning the rails between all of finance is structurally the same bet as one company owning the layer between the model and the agent. Whoever owns the pipe wins — the only fight is which pipe. Right, and the test is whether Tushar names a number the market's mispricing or just says 'the market is missing it.' When the guy who's long says 'the market is missing it,' you're hearing the position before the thesis. LawNext, with Bob Ambrogi:
When Anthropic launched Claude for Legal in May – more than 20 MCP connectors and a dozen practice-area plugins built specifically for legal work – it marked the company’s biggest and most explicit commitment yet to the legal industry.
This is the one that actually cleared a market bar this week — Claude for Legal got press, sure, but it also moved stock prices. Anthropic's first legal plugin briefly rattled Thomson Reuters, RELX, and Wolters Kluwer back in January. Right, and that's a falsifiable signal. Incumbents don't get spooked by a demo. The question is whether Bob Ambrogi pushes Pike on whether that move was justified, or lets him coast in launch mode. The May launch is what I keep coming back to — twenty-plus MCP connectors and a dozen practice-area plugins. That's practice-area infrastructure, way beyond a chatbot bolted onto contracts. Which is the funny part on today's rundown. Anthropic's the same shop warning open-source could torch the commercial AI model, and here's Pike selling a paid legal SKU with a dozen plugins. Existential threat on Monday, per-seat pricing on Tuesday. If Tech Podcast Podcast helps you keep up, take a second to subscribe or leave a review wherever you're listening. It really helps other curious tech listeners find the show.
Links to everything we talked about today are in the show notes, so if one story deserves a closer read, that's the place to start.
That's Tech Podcast Podcast for this Wednesday. This is a Lantern Podcast.