Dev Interrupted cover art
Podcast · 208 episodes

Dev Interrupted, page 2

by LinearB · English

Software itself is fundamentally changing. We explore the transition to agentic orchestration, vibe coding, and AI-native development, grounding the conversation in the principles that have always defined great engineering. On Tuesdays, we interview the founders, architects, and builders of the world’s most impactful tech to uncover the timeless engineering principles and strategies shaping the next era of development. And on Fridays, we drop an end-of-week roundup of the biggest news in AI and software, and what it actually means for your career, your craft, and your life as a developer. Subscribe to stay ahead of the next era of code.

All episodes, page 2

How LinearB helps Kraken find hidden bottlenecks across thousands of engineers | Nik SudanKraken’s Engineering Operations Lead Nik Sudan explains how to move AI pilots into production, then measure engineering health and ROI using LinearB (including its MCP server).30 Jun 2026 · 44 min · 16 chapters
The discernment horizon, loop-driven development, and a wizard’s very defensible pondAI’s real-world applications and limits, plus three Dev Interrupted-style reads: Steve Yegge’s “flat curve society” (models may become out of reach; “discernment horizon” where human grading caps…26 Jun 2026 · 29 min · 11 chapters
Your developers are the attack surface now and vibe coding as a vulnerability | Tanya JancaTanya Janca (SheHacksPurple) discusses OWASP Top 10 2025, why developers are now an “attack surface,” and how “vibe coding” (AI writing code without real review) creates sticky vulnerabilities.23 Jun 2026 · 46 min · 21 chapters
Microsoft’s wandering eyes, data labeling duties for senior devs at Meta, and prod is the new source codeThe episode covers (1) Anthropic’s Fable 5 access being disabled for foreign users after a reported jailbreak and the resulting shift toward model routing and smaller/self-hosted models; (2) claims…19 Jun 2026 · 38 min · 14 chapters
Your SDLC needs a productivity context engineLinear B founders discuss how AI codegen is reshaping the SDLC into an “ADLC,” and why teams need a productivity “context engine” to manage bottlenecks, quality gates, and ROI (not just adoption).16 Jun 2026 · 40 min · 17 chapters
How to harness your dragon with Fable, tech leaders turn to model routing, and coping with AI rockstarsBots and AI are reshaping the internet (over half of traffic), raising security and archiving concerns; the episode then focuses on Anthropic’s Fable 5 (a “Mythos” dragon model), token-maxing…12 Jun 2026 · 36 min · 10 chapters
All software is an optimization of tokens and time (and speed is still the moat) | AMD’s Anush ElangovanAgentic AI and open-source software engineering at scale, framed as “software is tokens and time” with “speed still the moat.” Discusses how AMD’s Rock ’em stack enables agents to rewrite and improve…9 Jun 2026 · 52 min · 18 chapters
Microsoft breaks free from OpenAI, using your harness to add drag instead of velocity, and the Linux built-ins you're sleeping onAI’s impact on software delivery and engineering trust, plus practical guidance on AI adoption.5 Jun 2026 · 27 min · 8 chapters
How to turn your 1000x engineer into a 10x everyone | LinkedIn’s Karthik RamgopalHow LinkedIn builds and scales “agentic” AI for engineers and other functions—turning a 1000x engineer into a 10x effect via durable context, memory, evals, and open standards (e.g., MCP).2 Jun 2026 · 53 min · 25 chapters
The cost of intelligence will never be this cheap again, the failure of intensive specs, and how bots disguise inefficient workflowsOpus 4.8/Claude model updates; moving AI inference and fine-tuning local to control rising API costs; AI data center networking/GPU architecture; avoiding “AI as a crutch” by fixing upstream workflow…29 May 2026 · 39 min · 15 chapters
Observability is your profit center now | Honeycomb’s Christine YenObservability as a profit/accelerant center in the AI-agent era; using telemetry to close the loop between customer impact, engineering decisions, and agentic development.26 May 2026 · 48 min · 24 chapters
What Google didn’t announce at I/O, defining “dark flow” and ignoring your first brain to build your second oneThe episode recaps Google I/O news and debates how AI agents should be integrated into real workflows.22 May 2026 · 30 min · 14 chapters
Android is the frontier for agents and other lessons from Google I/O | Matthew McCulloughMatthew McCullough (Google VP Android Dev Experiences) discusses Google I/O’s Android strategy for AI agents—“device is the frontier”—and how Android dev workflows are shifting from Q&A help to…19 May 2026 · 27 min · 10 chapters
Agents get their own AOL, Andrew gets published, and vibe coding is actually good?(1) AOL Agents Online, a nostalgic chat-room-style collaboration space for AI agents, with “ASL” reinterpreted as agent context (model, Windows context, capabilities, connected MCP servers).15 May 2026 · 41 min · 14 chapters
It’s Tuesday and your tech stack is obsolete (again). Now what? | Theory Venture’s Bryan BischofHow AI systems become “obsolete” fast, and what to do next—using conference track design, inference-layer “levers,” and agent engineering instead of hype.12 May 2026 · 56 min · 21 chapters
Goblins in prod, the messy middle of AI adoption, and everything is a harness nowThe episode covers “goblins” in LLM outputs, the messy middle of AI adoption, and why agent safety and permissions matter. Guests/hosts: Ben Lloyd Pearson and Andrew Ziegler (co-hosts).8 May 2026 · 31 min · 15 chapters
Teach the primitives or watch your competitor define them | Baseten’s Philip KielyPhilip Kiely (Base 10) discusses “Inference Engineering” and how AI education roles should teach developers to think about inference as a stack of primitives, enabling reliable, AI-native engineering…5 May 2026 · 36 min · 23 chapters
Tokenmaxxing scoreboards, the vegan LLM from before 1931, and 30% of the web is now AI-generatedThis Friday Deploy episode covers: (1) AI agents deleting production data when given an unscoped Railway token; the agent wiped a production DB and three months of backups in one prompt, then…1 May 2026 · 38 min · 16 chapters
Giving robots a brain | Intrinsic’s Brian GerkeyPhysical AI for robotics—how modern AI (especially neural networks/LLMs) plus better software makes industrial robots more capable in messy, variable environments.28 Apr 2026 · 52 min · 23 chapters
The harness is the showdown, the humans are the tool calls, and have you seen my Claude Code buddy?AI pricing “harness” chaos, how model choice and token limits change behavior; Claude Code/Anthropic pricing experiments; leaked Claude Code source patterns for production coding agents; and a Vercel…24 Apr 2026 · 37 min · 15 chapters
The best model for your team? You haven’t invented it yet. | Ai2’s Tim DettmersTim Dettmers argues that teams don’t need “industrial kitchen” compute to build strong AI coding agents.21 Apr 2026 · 45 min · 19 chapters
The self-authoring wiki, beating brain fry, and Obsidian as memory is a trapGoogle will punish websites that hijack the back button, framed as reducing agent “loops” and spam/vibe-coded web practices.17 Apr 2026 · 38 min · 15 chapters
The guardian in the machine | Wayfound’s Tatyana MamutAI agents need “guardian” supervision for business alignment, because agents are stochastic and can fail in ways single-turn testing won’t catch.14 Apr 2026 · 45 min · 23 chapters
Reading model benchmarks like a pro, Mythos is looming, and Claude talk caveman, save big tokenThe episode covers AI’s cybersecurity arms race, how to interpret AI benchmarks/scorecards, recent open frontier model releases, and a “caveman” output-compression plugin.10 Apr 2026 · 31 min · 13 chapters