AI-Driven Development with Olivia McVicker

January 19, 2026·39 min·3 clips

McVicker challenges the 'AI will write all the code' claim by asking whether it should — and why deploying unreviewed AI code is dangerous.

The frame is practical on purpose. Thomas Betts starts with AI-driven software development as a spectrum, from tools developers already know to newer capabilities that are still taking shape. Olivia McVicker sees the space as exciting, but the conversation stays close to everyday engineering work. The main example is an instructions file. Betts connects that habit to postmortems: a team finds a root cause, writes it down, and adds checks so the same failure is less likely to come back. The episode applies that loop to coding assistants. A fixed bug can become context for later code, not because the assistant has perfect memory, but because the team has told it what patterns matter. Cognitive load ends up being the main topic. A project can spell out known defects, stack-specific traps, and local style preferences before asking an assistant to write or review code. That makes the relationship a little more useful both ways. Better guidance for an AI agent can also push the human team to document practices it had been carrying around informally. Review is where the idea gets sharper. Betts imagines an agent reading his code and catching a bug because the instructions already describe that failure mode. McVicker treats that as part of where AI-driven development is heading. Her view puts the work less in one tool and more in a process built around documentation, feedback, and conventions. The ending stays provisional: models, integrations, and token limits can change fast enough that a bad experience may be worth revisiting later.

As heard by us

A grounded look at AI coding tools as team memory, code reviewer, and fast-changing development habit.

AI-driven development comes across here as a practical shift in team workflow, not a pitch for instant code generation. Thomas Betts and Olivia McVicker talk through the tools developers are already using, then move into coding assistants, agents, and the habits that make them…

Read the full review in PlayNext →

Why you'd press play

You want a grounded look at AI-driven software development, from current tooling to where it is headed.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from The InfoQ Podcast