Stay Sassy on AI budgets, code review, and build vs. buy
Anonymous tech writers Stay Sassy join Latent Space host swyx to discuss managing AI budgets on large teams, per-person token spend, build vs. buy decisions, and why code review becomes more — not less — important as AI coding tools proliferate.
Score breakdown
Engineering leaders and AI practitioners can use this discussion to frame internal conversations around token budget governance, code review rigor, and when to build versus buy AI tooling — practical concerns as AI-generated code becomes a larger share of production systems.
- 01Stay Sassy is an anonymous blog, Substack, and podcast about building and scaling technology businesses, run by two authors identified as Stay Sassy PM and Stay Sassy EM.
- 02The publication grew early through Hacker News front-page posts, then expanded to Substack and Twitter/X.
- 03The episode covers AI budgets and per-person token spend as management-level problems for large teams.
In this Latent Space episode, swyx hosts the anonymous duo behind Stay Sassy — a blog, Substack, and podcast focused on building and scaling technology businesses from early startup through growth stages. The guests, identified only as Stay Sassy PM and Stay Sassy EM, discuss how they grew an anonymous publication: early traction came from Hacker News front-page appearances, followed by a move to Substack for community building, and then Twitter/X for broader engagement. They note that seeing clusters of employees from the same top companies subscribe over time has been a particularly gratifying signal of organic word-of-mouth spread.
The bulk of the conversation centers on how engineering and product organizations are navigating AI tooling.
The bulk of the conversation centers on how engineering and product organizations are navigating AI tooling. Key themes include treating AI budgets and per-person token spend as genuine management problems, identifying which problems are well-suited to AI versus where hand-coding remains necessary, and how the software development lifecycle (SDLC) is changing. The guests frame the maturation of AI agents as a progression from "temps or interns" to full "employees," and discuss what larger or more established buyers should consider when deciding whether to build or buy AI products. A recurring point is that code review and reliability practices become more critical — not less — as AI-generated code becomes more prevalent, reflecting the episode's title thesis that the best engineers delete more code than they write.
Key facts
- 01Stay Sassy is an anonymous blog, Substack, and podcast about building and scaling technology businesses, run by two authors identified as Stay Sassy PM and Stay Sassy EM.
- 02The publication grew early through Hacker News front-page posts, then expanded to Substack and Twitter/X.
- 03The episode covers AI budgets and per-person token spend as management-level problems for large teams.
- 04Discussion topics include build vs. buy decisions for AI products targeting larger or established buyers.
- 05The guests frame AI agent maturity as a progression from 'temps or interns' to full 'employees.'
- 06Code review and reliability are argued to matter more, not less, in the age of AI coding tools.
- 07The episode also addresses executive augmentation with AI before individual contributors (ICs) and the boundary between AI automation and human decision-making.