No Figma. No Jira. No docs. How Gusto built a new product line with Claude Code | Eddie Kim (CTO)
How I AI
Jun 29
No Figma. No Jira. No docs. How Gusto built a new product line with Claude Code | Eddie Kim (CTO)
No Figma. No Jira. No docs. How Gusto built a new product line with Claude Code | Eddie Kim (CTO)
How I AI
Jun 29
Gusto CTO Eddie Kim details how he and a tiny team of five built a new AI product, Gusto Cofounder, from scratch in just 10 weeks. They achieved this by discarding nearly all traditional engineering processes, including meetings, documentation, and project management tools, in favor of a radical, code-first approach.
The team operated with a 'perma-Zoom' room, replacing standups and Slack threads, and used a 'trash-can method' where they would write and delete full pull requests as a way to make product decisions. The core tech stack was simple, using Cloudflare Workers and the Vercel AI SDK. A key outcome was that a designer with no engineering background became one of the top code contributors, ranking in the 94th percentile for PR throughput, by learning to prompt AI tools effectively and receiving fast, prioritized code reviews. Eddie emphasizes that leaders must write production code themselves to understand AI's real-world nuances, and that the low cost of building and discarding prototypes makes this approach viable for any team willing to break from convention.
00:00
00:00
Built in 10 weeks with 5 people, no meetings, no docs.
05:40
05:40
Long breaks often spark innovation.
08:33
08:33
Built in 10 weeks with no meetings, no tech specs, no Figma, no Jira.
12:51
12:51
Everyone acted as a PM.
14:38
14:38
Low cost of code makes discarding prototypes feasible
17:18
17:18
Building agents is not as intimidating as it seems
19:10
19:10
Ship a fake prototype, then chip away like marble.
22:04
22:04
Technical curiosity and team mentorship enable non-engineers to ship production code.
29:06
29:06
Gusto Cofounder integrates with existing data and supports multi-channel communication.
34:36
34:36
Act on ambitious ideas immediately rather than planning months ahead.
36:27
36:27
Write a failing eval, then fix the code to pass it
39:39
39:39
Keep the team small and skip the docs.
46:00
46:00
Hands-on coding reveals product insights
