Gusto's 10-week AI product sprint: A blueprint for scrappy GTM teams

Jun 29, 2026 · Lenny's Podcast
🎧 PodShort 52 min squeezed to 2 AI SprinklerAS Sales Tech New
Episode artwork
Eddie Kim
CTO and Co-founder at Gusto
Lenny's Podcast
52 min squeezed to 2
Full episode from Lenny's Podcast
Quotable Moments

We had no meetings, we had no text specs, we had no Figmas, we had no Jira board where we track stories or track work. We had nothing.

Eddie KimCTO and Co-founder at Gusto
00:00

I call this the trash can method of software engineering right now where you can actually trash all the code, start like a /v2 branch, and rebuild it from scratch and it's totally reasonable to do because the cost of the code is so low.

Claire VoProduct Leader and AI Obsessive
01:00

I think my advice to leaders is like actually get hands-on in building like production code. Don't just... I think it is an important first step to build a prototype and come to your team and be like, 'Look, like this is actually feasible and possible.'

Eddie KimCTO and Co-founder at Gusto
45:35
Key Insights
  • Gusto's co-founder product was built by a lean team of 5 people (3 engineers, 1 designer, and Eddie Kim himself) in just 10 weeks, from initial idea to a tier-1 launch.
  • The development process was characterized by an extreme lack of traditional overhead: zero meetings, no text specs, no Figmas, no Jira boards, and no stand-ups or retrospectives. The only structured tool was a 24/7 'Perm-Zoom' room.
  • The technical stack for the product was surprisingly simple, utilizing Cloudflare Workers for the actual agent loop and Vercel AI SDK, with no other complex harnesses.
  • The concept for the product originated during a 5-hour layover, where Eddie Kim prototyped the core idea using Claude Code, which was then iterated upon by his small team.
  • The 'trash can method' of software engineering, where the cost of writing code is so low that teams are comfortable deleting entire features or rebuilding from scratch, enables rapid iteration and minimizes commitment to early solutions.
  • Product decisions were made collaboratively during real-time code reviews within the Perm-Zoom room, with a willingness to delete features immediately if they didn't make sense, due to the low cost of iteration.
  • Designers on the team were empowered to ship directly to production by creating 'faked' frontend experiences, which engineers would then progressively connect to real backend logic, allowing for continuous iteration and improvement of the user experience.
  • Leaders should actively participate in hands-on building of production code, not just prototypes, to truly understand the nuances of the work and to demonstrate the feasibility of rapidly developing ambitious AI-powered products.
Metrics Mentioned
  • 5 people (3 engineers, 1 designer, 1 CTO/engineer) (The size of the team that built Gusto's co-founder product.)
  • 10 weeks (The duration from initial idea to tier-1 launch for the Gusto co-founder product.)
  • Over 1000 people (The size of Gusto's overall R&D organization.)
  • 94th percentile (Katie, the designer, ranked 94th percentile in throughput (PRs landed in production) across Gusto's entire R&D organization.)
  • 9 minutes (The median PR review time for the Gusto co-founder development team.)

RevBots.ai View:

  • AI Sprinkler teams can adopt Gusto's 'zero-process' approach to accelerate MVP development.
  • The 'trash can method' reduces technical debt fears common in SaaS Hoarder orgs with bloated stacks.
  • ARM-stage companies will institutionalize this rapid iteration as core to product-led GTM motions.
🎧Full Episode:Lenny's Podcast →