Dry Run

Austin Campbell, founder of Zero Knowledge and adjunct professor at NYU Stern, has spent his career in risk and structured finance. In this episode, he dives into the ins and outs of the AI buildout, especially how it is financed, who's doing the financing, the role of debt, and how the AI boom compares to past bubbles, from the dot-com era to 2008.

03:39 Components of the physical buildout

07:41 Capex spent so far vs. future commitments

10:13 Why Big Tech buybacks are dropping

13:18 How SPVs are structured and good vs. bad reasons to use them

20:13 AI makes money, but can it pay back the capex?20:59 Who takes the hit if demand falls short

24:33 How today's SPVs compare to 2008

32:25 Minsky's three stages of a bubble, applied to AI Catch Dry Run wherever you listen to podcasts:

Spotify: https://open.spotify.com/show/57Re2SH8LhE99F5RnFHomZ

Apple Podcasts: https://podcasts.apple.com/de/podcast/dry-run/id6814644273

YouTube: https://www.youtube.com/@TheDryRunPodX: https://x.com/TheDryRunPod

Catch Austin & the hosts on X:

Austin Campbell https://x.com/austincampbell

Eshita https://x.com/eshita

Khushi https://x.com/khushii_w

Naman Kapasi, CEO and co-founder of Nirvana, joins us on the Dry Run to talk about why data quality, not intelligence, is what's holding general purpose robotics back.

02:05 LLMs to VLAs to world models

07:39 Playing air hockey with a robot

08:34 The bottleneck isn't perception, planning, or control

10:37 The tiers of robotics data quality

13:48 Scale AI vs Mercor

18:01 Nobody knows how long a humanoid lasts

23:52 Five fingered hands vs giving robots tools

27:24 At 80% nobody cares, at 99.9% it's useful

29:28 A robot can't tell if an avocado is ripe

30:09 The US just banned imported humanoids

Everett, Head of GTM Engineering at Clay, joins Dry Run for a conversation on the rise of GTM engineering and what it changes about sales.

Timestamps:

06:00 - Why the old outbound motion broke

09:51 - GTM loops: data > agents > action > feedback

12:46 - The “golden list” every founder should build

14:04 - Good outbound means sending fewer emails

16:24 - The campaign that booked ~60% of targets

19:48 - How other industries leapfrog into modern GTM

22:17 - The GTM mistake founders make early

25:19 - What sales agents will automate first

29:06 - Why GTM engineering is an index on AI capabilities

The internet has had a status code reserved for payments since the 1990s. Nobody could implement it until AI agents needed wallets.Lincoln Murr, AI products at Coinbase, joins Eshita and Khushi to answer an important question: how do agentic payments work?We go from buying a VPN with Bitcoin at 12 years old to whether agents will run their own businesses.00:31 Trading a Target gift card for Bitcoin at age 1201:34 What x402 actually is05:23 How AWS, Vercel, and Cloudflare legitimized agentic payments06:45 Why every previous attempt at 402 failed08:36 Agents can't get bank accounts09:12 Chargebacks, fraud, and reputation systems11:40 Agentic payments as a revenue driver for AWS14:46 The 2020s shift from mobile to agents16:16 Solopreneurs running companies with fleets of agents17:45 Why every company will want its own settlement layer20:15 AI Advisors: regulated financial advice at scale24:41 Asking an AI for a delta neutral basis trade29:28 Agents buying premium data to make better trades

Simon Corry has spent over two decades in product design, from the Royal Mail to WeTransfer to Ramp, and he's now at Basis. In this episode he joins us to talk about his design journey, how he hires for spiky people over polished portfolios, and what he learned building a multiplayer game solo with AI agents.

We get into the collapse of design, engineering, and product into a single role, why one of his agents started making its own decisions (and named its own game engine), AI brain rot and untested code slipping into production, and how Ramp Labs became a blueprint for how companies should experiment.

Follow Simon's game dev and agent-building journey here:

https://x.com/simoncorry/status/2062229653746577770

00:53 From call center manager to accidental designer

03:39 Why he's the worst employee you've ever hired

10:21 How he hires for spiky, interesting people

13:51 Shipping a multiplayer game by passing infra to Claude

16:54 How one of his agents decided to build a game engine

20:06 AI brain rot and untested agent code in production

23:52 Design, engineering, and product collapsing into one

26:29 Ramp Labs as the blueprint for experimentation

31:08 Building out a neolab at Basis

32:19 New media and testing the bleeding edge

Gap, founder and CEO of Suby, joins joins Khushi and Eshita on the Dry Run to talk about who really moves your money when you swipe a card, and where stablecoins fit into the stack.We go from a $40K NFT checkout problem at Louis Vuitton to why stablecoins transform global money movement.00:25 Reselling sneakers to running crypto at Louis Vuitton05:27 The $40K NFT chargeback problem08:47 The four companies behind every card swipe12:23 Why Stripe became a bank15:59 Uncapped US interchange vs Europe's 0.2% cap18:14 Where stablecoins actually matter21:59 Why Stripe struggles in Latam and India23:49 Tempo, agentic payments, and nano payments28:04 Neobanks don't hold your money36:07 Buy vs build: Bridge, Privy, BVNK39:28 Circle as fintech's plumbing40:43 What Gap is building at Suby

Amy Zhao from Initiative for Cryptocurrencies and Contracts (IC3) to break down her team's research on AI trading platforms that hit $3B+ in combined market cap.

We go from how users lost $191M to why most agents aren't trading autonomously yet, and what a maturity framework for investment-grade agents looks like.

0:40 The motivation behind studying AI trading agents5:36 Why Eliza OS and Virtuals captured billions10:22 Early market or something more predatory?14:49 Tradeoff between transparency and profitability18:10 The maturity framework for trading agents23:47 How does this apply beyond DeFi trading?26:08 The most unexpected finding from the research

Nitya, founder and CEO of Para, joins us to discuss how embedded wallets are reshaping the crypto user experience, and whether the traditional wallet model still makes sense.

We break down why wallets exist on blockchains, how they've evolved from 2017 to 2026, and how wallet infrastructure will need to adapt for a world of agentic payments and AI agents. We also explore product development and prioritization, and how the team at Para thinks about building core crypto infrastructure for the long term.

Topics covered:

03:14 The user journey into crypto

05:13 The evolution of wallets

09:14 Wallets vs. accounts

11:09 What makes one wallet better than another?

13:21 Should wallets be unopinionated infrastructure?

17:11 Wallets for agentic payments

Michael Blau is a former a16z crypto investor and the founder of Drip, a platform that lets agents pay creators directly for paywalled content.

We talk about why micropayments failed for humans but work for agents, how the dev stack for agentic commerce works, and what the internet looks like when your agent has a budget and is buying on your behalf.

We also get into why crypto creator monetization failed the first time, why podcasts are the next media type for pay-per-use, and how independent publishers are caught between AI pulling information open and paywalls pushing it closed.

Dan Bricklin, co-creator of VisiCalc and builder joins us to talk about how people adopt new tools.

We go from how VisiCalc launched on the Apple II into a world where few knew what a computer was to why he thinks most AI tools haven't found their real interface yet.

02:19 How the first electronic spreadsheet was made

07:05 People buying hardware just to run the software

08:13 Why the spreadsheet had to be 100x easier

18:19 Why Excel still can't be replaced

20:10 How people adopt new tools

27:13 VisiCalc made accountants better, not obsolete

36:14 The cornucopia of the commons

44:14 Do we want chauffeurs or tools

46:11 Advice for builders