Every company training voice models starts out by scraping YouTube for audio data. But the YouTube well is dry. Labs all need high-fidelity audio to scale. @usepanels is Scale AI for audio data. Congrats on the launch, @wenk_aaron & @jasonle_ ! 🚀

How you do anything is how you do everything. Are you in the details? Do you have taste? I want to be surrounded people for whom this is all a "hell yes." Those are basically the only founders I want to fund.

Panels (@usepanels) is an audio data platform that delivers large-scale, richly labeled speech datasets from vetted, diverse contributors to power the training and evaluation of foundational voice models. Congrats on the launch, @wenk_aaron & @jasonle_!

.@RoundtableHQ_'s Proof of Human uses behavioral biometrics to stop bots and AI spam. With 87% accuracy (vs. Google’s 69% and Cloudflare’s 33%), it gives you frictionless, real-time authentication with a one-line API. Congrats on the launch @_magrawal & @mdahardy!

If you're running a startup, sell to startups if you can. They don't have as much money, but they decide quickly, and even more importantly, they have a better influence on your product.

1 year ago I was living in Germany and broke. We pitched 50 VCs for $500k Pre-seed and all of them said no. One year later, I moved to SF and raised $4.7M from Nexus, YC and Paul Graham with the round done in less than a week. The YC effect:)

NEW: Zapier Agents now work like a team. Roles, handoffs, and closing the loop. What's new: Agent-to-agent calling: Your AI agents now pass work like a relay team, each handling the step they’re best at (no human handoffs needed) Copilot: Describe your workflow in plain language, and Copilot maps the workflow and builds your full agent team, start to finish. Live knowledge sources: Your agents now pull live data straight from Box, Dropbox & Google Drive, always staying up-to-date. Template sharing: Turn your best agents into @Zapier community templates. Share wins, showcase use cases, and turn your best builds into content. These specialized agents now collaborate to run full workflows (sales, support, marketing, and more), not just isolated tasks.

When I became a founder, I underestimated how hard it would be and how much it would change me. One thing I learned fast: ambitious people can do extraordinary things together. 🧵

In one of my favourite Cheeky Pint episodes to date, @brian_armstrong brings us behind the scenes at @coinbase: their battles against North Korean hackers, war stories from breakneck early scaling, confronting people who won’t use AI to code, Coinbase becoming people’s primary financial account, and how banks are now embracing crypto. Subscribe on Spotify, Apple Podcasts, and YouTube for future Cheeky Pint episodes. Timestamps 00:00 Intro 01:14 How Coinbase won the exchange market 06:22 Losing money on every bitcoin purchase 10:54 North Korean hackers 14:54 The Everything Exchange 17:39 Crypto will be bigger than gold 19:22 Stablecoin adoption 23:28 Jamie Dimon and big banks on crypto 26:44 Coinbase as your primary financial account 27:53 Neobanks 29:41 Bitcoin going to $1 million 33:16 Crypto in an investment portfolio 34:43 GENIUS Act 37:10 Electing a pro-crypto Congress 45:24 Reforming accredited investor rules 48:33 Squashing scams 50:11 Bitcoin preserving the American experiment 57:09 Balaji Srinivasan 01:01:05 The mission-first company announcement 01:05:50 How does Coinbase focus? 01:07:44 Venture bets and Brian vetoing USDC 01:12:02 Brian’s AI coding mandate 01:15:13 Advice for Stripe

IN NEWS: Pylon raises $31M Series B (co-led by @a16z & @BainCapVC). Pylon is going after the B2B slice of support, not the simpler, transactional B2C queues. “In consumer you have very simple transactional support… In B2B you don’t just have customer support, you also have customer success, solutions, professional services, account management.” –@marty_kausas (Co-Founder & CEO) Tickets need rich account context (CRM notes, call recordings, implementation plans), and routing spans multiple post-sales teams. “One way to think about us strategically is we’re building something similar to Rippling, but for the post-sales use case.”