Katyella

Episode ten: Jason Patel splits AI builds into planned projects and "copy and customize", where the tokens disappear fast. From there: Linear versus markdown files, giving each sub-agent its own Linux user, why agents don't need email, and whether engineers will make good team leads of agents. Matt and Jason compare how much access to give Claude, ask who clicks past an AI summary, and talk LLM tutors, messy government data, and hallucinations versus mistakes. Plus Matt's German learning app and the AI he let buy him pants.

Episode nine: Jason Patel is back, now as a regular co-host. Matt has been designing with Paper, a design tool his AI can drive, and Jason asks whether AI-generated design is good enough to ship. From there: AI features nobody asked for, whether it's all for the valuation, agents that forget and harnesses that lock you in, why engineers now have the product manager's spec problem, and what "correct" even means in back-end design. Matt shows two teams of AI agents pitching him designs, narrated by AI voices, and Jason explains why AI works best in English.

Episode eight, with Jason Patel: exited CTO of Ensighten, now building his own side project in Chicago, and someone Matt knows from enterprise engineering. A skeptical, experienced-engineer look at the AI rush: always-on agent servers versus a Lambda and a few API calls, why multi-agent orchestration pays the same tax as microservices, whether small businesses can really afford AI enablement agencies, and Jason's case that a lot of it is just standard automation. Plus the Claude Code fix that cleared one failing Docker pull and left another one further down the same script.

Episode six, with Chase Ripplinger: software developer, 10+ years across startups, blockchain, and finance, and a former colleague of Matt's. He writes zero code by hand now. Running AI-driven upgrades when Copilot is the only approved tool, a custom harness that does the code side of a Spring Boot 4 upgrade in about four hours, API tests over unit tests, and why proving the code is correct is the real bottleneck. Plus Home Assistant dodging peak electricity rates, a local LLM on a Mac mini, and the AI assistant that deals with Matt's tax people for him.

First guest episode: Ben Sun of Meadow joins from Berlin. Pirates vs the Navy on startup life, whether agile survives AI-speed development, review bottlenecks, bigger tickets, multi-agent testing, and what managed infrastructure is still worth. Plus the continuing saga of Nina not charging for her fiction app.

Matt and Nina sort the week's tool hype from the useful parts: multi-agent coding in Amp and Codex, model routers, and classifiers that don't need an LLM. Plus live-coding interviews in the AI era, big tech layoffs and golden handcuffs, Nina's collaborative writing app, and the case for finally charging for it.

The week the AI discourse turned dark: a possible coming attack on open source, Primeagen on protecting your energy, and whether letting agents make all the decisions causes skill regression. Plus legacy migration as the consulting angle AI de-risks, and why Nina built apps for years without ever sharing one.

Nina migrates Matt's podcast search app from Next.js and Python to her first-ever Java deploy, six years into a software career. Plus the economics of running agents on a subscription versus an API, multi-agent pipelines that pass JSON instead of prose, why AI job interviews feel terrible, and how episode 1 got its title.

Matt and Nina introduce themselves and get into what AI has done to building software: Nina ships a desktop audio tool that would have taken a team, Matt tries the cloud-agent chief of staff trend, and they argue about whether the curve is flattening and whether the silver bullet that kills programmers actually exists.