Engineering Unblocked

The episode offers substantial, concrete insights about AI-assisted coding that challenge common practices: distracted agents vs. focused agents, compliance bias, non-determinism exploitation, refactoring as a superpower with AI, and the critical role of mental model maintenance. The guest provides specific examples (the little committer, 180-slide processing, Agile 2025 case study) rather than abstract theory. However, some segments contain extended throat-clearing and repetition that dilutes the overall density.

“I saw what a focused agent looks like. I think many people are working with an anti pattern of distracted agent, which is basically you're like, hey AI, here's my rules. And also do this, do this, do this, do this, do this.”

“one pass with AI doesn't work. It doesn't actually produce good results for anything more complex than a simple, simple question.”

The guest introduces genuinely fresh frameworks not widely circulated: the 'focused vs. distracted agent' distinction, 'compliance bias' as a specific AI challenge, 'feedback flip' pattern, the concept of AI as having limited 'focus' rather than just context, and the dual-universe framing of future AI development (autonomous systems vs. human-in-the-loop). These are first-principles observations grounded in hands-on experience rather than recycled productivity frameworks.

“there's a concept of focus in the eye, right? So it's not just context.”

“compliance bias. Right. So it's trained deliberately to, to say yes, to say you're amazing. Uh, you're right.”

Lada Kessler is a hands-on practitioner with deep expertise in both software craftsmanship (TDD, refactoring, domain-driven design) and year-long, intensive AI coding experience. She's worked as a technical coach at Logic 2020 (enterprise consulting) and contributed to real-world solutions (Agile 2025 impact). However, she is not a household name, CTO-level operator at scale, or someone with multi-year track record managing AI in large production systems; her credibility rests on demonstrated hands-on exploration rather than seasoned organizational leadership.

“I'm a software developer and I think of myself as a crafter...my roots are really in like the deep Agile, like the early Agile...extremely programming, right? So the 2 TDD refactoring domain driven design”

“This year I've been just doing so much hands on coding and writing agents and whatnot with it.”

The transcript includes concrete examples: a focused 'little committer' agent vs. a distracted main agent, the 180-slide markdown processing task, GPT Codex model failure, a working system of 50 microservices redesigned in a month, and the Agile 2025 case study solving two real problems for abuse prevention funding. However, many claims lack numbers (What were the exact metrics? How much faster was the committer? What were the cost savings for the microservices client?). Financial and time-based specificity is sparse; most evidence is narrative rather than quantified.

“I gave it a small task. So I gave it. Okay, it wasn't a small task. To be fair. I have 180 slides on the first talk and I'm like, look at every slide and make me a textual representation in markdown”

“I've been dropped into system of 50 microservices. And that I didn't know, right? In a month I emerged from that, uh, having changed the architecture of 10 of them in a better way, plus delivering the feature they care about and saving them a few millions of dollars”

Rebecca Murphy asks thoughtful, substantive follow-up questions ('How can you create a system where AI can be beneficial?', 'Are best practices still relevant?', 'Why did you raise an eyebrow at documentation?') and shows genuine technical knowledge by building on answers (GitHub Actions for documentation updates). However, the host rarely pushes back or challenges claims directly; she affirms Lada's framings ('I really agree with that,' 'yes, everything you just said is correct'). The conversation lacks productive tension - no moments where Rebecca forces deeper justification or exposes contradictions. Most exchanges feel confirmatory rather than interrogative.

“So how do you create a system where AI can be beneficial and not detrimental, given those challenges?”

“And I find myself writing more software than I've written in years.”