Latent Space ran a hackathon to clone a $43,000-a-year SaaS in a weekend
Most 'SaaS is dead' talk stays theoretical. swyx and the Latent Space podcast tested it: they took a real $43,000-a-year SaaS they use and ran a global hackathon to clone it in a weekend, judged with both AI and human evals, with any model and any harness allowed. Seventy-two submissions and about 100 billion tokens later, they published the data, what's still hard about killing a non-trivial mid-tier SaaS, and interviews with the winners of the $10,000 Kill My SaaS contest. A proper boss fight, with a scoreboard.
🆕 How to Kill Your SaaS
A lot of people talk about the death of SaaS but mostly in theoretical terms. We took a real $43,000/year SaaS we use and launched a global hackathon to clone it in a weekend, with both AI and REAL HUMAN evals, using any models any harness.
72 submissions, a lot of manual evals and ~100B tokens later, here are the results
In this special @latentspacepod edition we share the data collected, what's still hard about killing nontrivial, midtier SaaS, and reveal the winners of the $10,000 Kill My SaaS contest with qualitative interviews!
— swyx 🗽 @aiDotEngineer, read the original on XMore SaaS obituaries
- David Kobia hit a broken support link and vibe coded a SendGrid alternative by Sunday night
- Hunter Bohm cancelled 5 subscriptions in two weeks with Claude Code, about $134 a month
- Kelly C. built a half-decent CRM in about the time it took to research and price CRMs
- Noah Kagan says Claude Max killed $15,788 a year of software, Carta and Zapier included
- Pawan cancelled $1,700/month construction software after bots rebuilt it on QuickBooks