Different Teams, Different Rules: What Senior Engineers Actually Let AI Do (Scaling Tech Podcast Episode 65)

We talk about “AI-assisted development” as though everyone is doing the same thing. But no two engineering teams are exactly alike. The culture is different. The processes are different. The constraints are different. And so is AI agent autonomy: how much decision-making power each engineer is actually comfortable handing over.

Using AI on a small team where two engineers own nearly everything is very different from using AI when a seemingly simple change touches backend services, mobile applications, downstream products, QA teams, third-party vendors, security processes, and then other people’s code.

In this episode of the Scaling Tech Podcast, guest host AgilityFeat & WebRTC.ventures COO Mariana Lopez, brings together three senior software engineers from her team working in very different environments for a comparison of what AI-assisted development actually looks like for them. Not theoretically, but what they’re actually comfortable letting AI do.

  1. Tahir Gölge works on a 13-14 person team building a complex product across backend, mobile, QA, and third-party vendors, now preparing for production after years as a research project. 
  2. Andrea Phillips works on a small team inside a large, healthcare-regulated organization with established frameworks and architecture already in place. 
  3. Justin Williams works on a small team with significant ownership and autonomy, on a codebase he’s known since before AI tooling existed.

If your team is figuring out how much autonomous AI coding to allow and what AI guardrails actually matter at your scale, this Scaling Tech Podcast episode is essential listening.

This episode is brought to you by AgilityFeat, providers of technical staffing, software development and AI integration services from Latin America. Learn more at agilityfeat.com.

Listen or watch on Spotify
Listen on Apple Podcasts

Chapters

  1. 00:00 Intro & Sponsor: AgilityFeat
  2. 02:16 Three Engineers, Three Very Different AI Environments
  3. 09:49 AI Agent Autonomy: How Much Work Do You Actually Hand Over?
  4. 15:26 Guardrails for AI-Generated Code
  5. 21:57 Giving AI Agents the Right Context
  6. 27:17 AI Code Review: How Engineers Verify AI-Generated Work
  7. 34:27 Multiple AI Agents: More Isn’t Always Better
  8. 36:36 What Gets Harder When Coding Gets Easier
  9. 40:06 Keep, Kill, Start: Practices for Agentic Development
  10. 47:57 Outro