English-language podcasts

How I AI

How I fight token fraud as a solo founder

40 min

Listen through the publisher, then discuss the episode here or arrange an in-person Roundtable. Your city determines where to meet.

Roundtables are designed for groups of four, with a shared subject and room for everyone to speak. Come for discussion, with no dating expectation.

Today I’m walking you through a problem that has cost me tens of thousands of dollars over the past year and eaten almost all of my engineering energy: token fraud. I’m the solo founder of ChatPRD, a bootstrapped AI platform for product managers, and fraudsters have been farming our free trials, reselling promo codes, wrapping the product in relays, and installing a custom MCP that ran as a 24/7 compute worker. I’ll walk through the six abuse patterns I’ve seen, the business and technical controls I put in place, the Jev loop that classifies fraud in real time, and why I finally killed free trials for good. If you sell AI, you sell compute. What you’ll learn: What token fraud is and why AI SaaS is uniquely exposed The six attack patterns I’ve seen at ChatPRD Why I finally turned off free trials for good The Roblox and Fortnite maps I accidentally funded Business controls that actually slow fraud down How I use Jev to classify fraud in real time Why your AI support desk can become an attack vector What “if you sell AI, you sell compute” actually means — Brought to you by: Auros (https://aurosglobal.com/howiai)—The human network behind AI — In this episode, we cover: (00:00) Token fraud and what it cost me (05:25) Six abuse patterns (11:24) Four kinds of abusers (13:45) Business controls (21:27) Technical defenses (30:25) Bots that watch the bots (35:30) Final takeaways — Tools referenced: • Stripe Radar: https://stripe.com/radar • Vercel AI Gateway: https://vercel.com/docs/ai-gateway • Vercel Bot ID: https://vercel.com/docs/botid • Vercel Firewall: https://vercel.com/docs/vercel-firewall • Clerk: https://clerk.com/ • Codex: https://openai.com/codex • Jev (TypeSafe AI): https://docs.typesafe.ai/introduction/quickstart — Other references: • Claire Vo on token fraud at ChatPRD: https://x.com/clairevo/status/2119985786045227197 • Gareth Wham (Atlassian) on fighting AI abuse: https://x.com/gareth_w/status/2120226872177095114 • FBI Internet Crime Complaint Center (IC3): https://www.ic3.gov (https://www.ic3.gov/) • Simon Willison’s pelican riding a bicycle: https://simonwillison.net/tags/pelican-riding-a-bicycle/ • How I AI: I’m using Jev more than Opus 5.5 or GPT-6: https://www.youtube.com/watch?v=-KIBgpGA_XI — Where to find Claire Vo: ChatPRD: https://www.chatprd.ai/ Website: https://clairevo.com/ LinkedIn: https://www.linkedin.com/in/clairevo/ X: https://x.com/clairevo — Production and marketing by https://penname.co/.

Episode information is provided by the publisher. Listening takes place on the publisher’s service.

Listen at the source ↗

Featured places to meet in Palo Alto, CA

Explore these places and arrange a conversation through ThinkMatch.

City guide for Palo Alto, CA

More from How I AI

How I AI

Claude Code + Obsidian: My family’s AI brain | Scott Farrar

35 min

Scott Farrar is a math educator and parent who got tired of AI chats that drift and forget. So he built a household brain in Claude Code: a folder of Markdown files, viewed in Obsidian, that knows his appliances, warranties, insurance documents, and everything on his family’s to-do list. In this episode, Scott shows how he uses it to shop for car insurance, plan repairs, share the system with his wife, and put priority tasks and recent wins on a wall display. We also talk about AI and math learning, and why he wants AI to keep kids doing the thinking. What you’ll learn: Why he moved his home life from AI chat to Claude Code What Claude told him to photograph first How AI made him a sharper insurance shopper The to-do file he keeps coming back to How he and his wife share one household brain Why his wall display has a “recent wins” section How a flipped tutor gets kids to teach the AI — Brought to you by: Jira Product Discovery (https://atlassian.com/howiai)—Prioritize with insights, bu

Discuss this episode

How I AI

Claude Fable 5.2 review: fast, creative, and finally not annoying

25 min

Anthropic just released Fable 5.2, their new frontier model, and I spent two days actually building things with it before recording this. I made an animated marketing video from a single tweet, assembled a pile of unboxed furniture using only blurry floor photos, and extended our CXO brand sprites into a full bug-defense game complete with a trailer. In this episode I walk through what impressed me, what fell down, and how I’m thinking about routing tasks between Fable 5.2 and Opus 5.5 going forward. What you’ll learn: What Anthropic launched with Fable 5.2 and Mythos 5.2, and when to reach for Fable instead of Opus 5.5 My pros and cons after a couple of days of hands-on testing How Fable 5.2 turned a tweet and a brand book into an animated video that doesn’t look like slides The geometry test every other model failed: assembly instructions for flat-pack furniture with no manual How it handled a real re-architecture of ChatPRD’s product intelligence pipeline Why I still wouldn’t let it

Discuss this episode

How I AI

How this founder built a better Duolingo with Codex | William Imoh

39 min

William Imoh is the founder of Hackmamba, a developer marketing company, and the creator of Mural, a real-time voice app he built to learn Norwegian after moving from Dubai to Norway. He has a chemical engineering degree, taught himself to code, and spent four years as a product manager, but he had never written a line of Swift. In this episode, William shows how he used Codex and GPT-Live to build Mural natively for iOS and Android, then walks through the Lord of the Rings–themed Grok Bot “fellowship” that helps him run a team of about 25 as a solo founder. What you’ll learn: The design prompt that made Mural look nothing like any other language app Why 1,000 days of Duolingo left him unable to hold a conversation in Norwegian What GPT-Live-1 costs per session and the latency threshold that makes or breaks a voice tutor How he assigned every solo founder role to a Lord of the Rings character in Grok Why he gave his agents access to everything across 15 open tabs, except one thing What

Discuss this episode

Community Poll

Communities should test major changes on a small scale before making them permanent.

Share your view, then see how readers voted. No sign-in required. This poll is shared across stories.

Do you agree or disagree?

Choose one answer

Common Quandary

A Common Quandary to discuss

Another conversation to explore, separate from this news story.

A stranger rings up fancy apples as the cheap kind. Do you mind your own business?

At self-checkout, the shopper beside you labels expensive organic apples as the cheapest variety. An employee is distracted nearby. The difference is only a few dollars, the store is a huge chain, and the shopper notices that you noticed.

When does witnessing a small cheat make it any of your business?

Explore this quandary