Oct 30, 2026 · 31 min
Spawn turns ten people into an AI software factory
How the Most AI-Pilled Company in the World Works
Spawn’s operating model tests how far a small company can scale software production with AI agents without surrendering human judgment.
- 1Spawn uses an always-on fleet of AI agents to generate an extraordinary volume of software changes.
- 2The company’s AI-native model depends on human ambition, taste, and strategic direction rather than eliminating them.
- 3Spawn’s two years of experience offer practical lessons for businesses deciding how deeply to integrate AI into their operations.
Don't miss
The episode’s clearest case study is Spawn’s ten-person team using an always-on AI-agent fleet to generate tens of thousands of software changes each month.
The brief
Nathaniel Whittemore opens with a company where AI agents handle nearly everything except long-term vision and direction, setting up a test of what “AI native” really means.
The focus turns to Spawn, a ten-person company that has spent two years building an always-on fleet of AI agents and using it to produce an extraordinary volume of software changes.
Spawn’s model is not presented as automation without people: human ambition, taste, and strategic direction remain the scarce inputs that determine what the agents should build.
The discussion extracts practical lessons for other businesses while placing Spawn’s approach alongside major AI industry developments and the infrastructure required to support it.
What was said on this episode
14 statements · 10 positive · 2 negative · 2 neutral
Spawn agents handle nearly all work except long-term vision and direction
“agents were doing basically everything other than setting long-term vision and direction”
Listen at 0:03
Spawn’s agent fleet can push 77,000 changes monthly
“agentic fleets that are able to push 77,000 changes a month”
Listen at 0:12
Weil’s corporate records could be worth tens of millions to a suitable buyer
“Weil's corporate records, which would have historically been valued at zero, could be worth tens of millions of dollars to the right buyer”
Listen at 3:10
Crosby could use Weil’s data and brand to build an agentic white-shoe law firm
“Crosby could have the top-quality data and brand required to make a run at building the first agentic white-shoe law firm”
Listen at 3:41
Crosby’s purchase would be a huge unprecedented bet on agentic firms
“it will be a huge and unprecedented bet on the agentic firm”
Listen at 5:11
Spawn’s ten-person team ships a product built on a large agentic stack
“his 10-man team is now shipping an impressive product built on a huge agentic stack”
Listen at 16:24
Spawn spent $3.1 million on tokens, thirty times its payroll spending
“Spawn spent $3.1 million on tokens last month, 30 times what they spent on payroll”
Listen at 18:11
Spawn’s fleet shipped 77,000 code changes in one month
“The fleet shipped 77,000 code changes last month, an average of 108 an hour”
Listen at 18:19
Spawn’s agents work directly in the codebase without pull requests
“Spawn killed pull requests and just had the agents work directly in the code base”
Listen at 21:11
Spawn operates somewhat like self-healing software
“This makes Spawn somewhat like self-healing software”
Listen at 22:49
Spawn’s AI judges made over 11,000 decisions in eight days
“The first AI judges came online on October 6th and made more than 11,000 decisions in the first eight days”
Listen at 24:35
AI judges increased Spawn’s daily changes from about 1,000 to over 7,000
“Before implementing the judges, the fleet shipped around 1,000 changes per day. After the judges had scaled up, the fleet shipped over 7,000”
Listen at 24:56
Spawn’s operating system is difficult to build from scratch
“this is the type of system That's hard enough to build when you're building it from scratch”
Listen at 28:14
Ambition and long-term perspectives do not come naturally to agents
“understanding that ambition and long horizon views do not come natural to agents”
Listen at 30:15
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.