Most important take away

The most powerful prompting technique is ending every AI interaction with two questions: “What am I missing?” and “What do you recommend?” These two additions leverage AI’s vast checklist of considerations and will dramatically improve the quality of output. Beyond prompting, the biggest wins from AI come not from replacing developers but from targeting the most painful, time-consuming tasks first to build organizational buy-in.

Summary

Rob Conery, author of The Imposter’s Handbook and former Microsoft VS Code content creator, shares his current approach to AI consulting with Fortune 500 companies. His core insight is that AI adoption is fundamentally a human and organizational challenge, not a technical one.

Actionable Insights:

• Two magic prompting questions: Always end prompts with “What am I not thinking about?” and “What do you recommend?” This forces the AI to surface considerations you haven’t hit yet and provide opinionated guidance. Even if the AI can’t implement perfectly, it plants ideas you can act on.

• Start with the most painful problem: When introducing AI to a skeptical team, find the most time-consuming, boring task and automate it first. Rob turned a 6-week ETL process into a 90-minute task, which instantly converted two vocal AI skeptics into enthusiasts.

• Use AI as an architecture sparring partner: Treat AI like a senior architect — describe your thinking, ask if the approach is sound, and get recommendations on tooling. Nine times out of ten it will confirm your direction but suggest a better tool or eliminate unnecessary complexity.

• Don’t fire people — redeploy them: When AI increases productivity, resist the knee-jerk to downsize. Instead, attack the massive backlog of tech debt, security scanning, bug hunting, and features that never made the priority cut. Anthropic’s work finding hundreds of bugs in Firefox with AI is a model for what every company should do with their own codebase.

• Don’t try to convince skeptics — let results speak: Rather than arguing about whether AI is a fad, demonstrate value on real problems. Tell skeptics “I’ll be here when you convince yourself in the next few months” and give them a few things to try on their own.

• Subject matter experts are already building their own tools: Non-developers are using AI (Claude, etc.) to build their own applications to bypass IT bottlenecks. This signals that dev teams need to restructure hierarchies and have developers work directly with domain experts as AI trainers and guides.

• Everyone becomes a generalist again during disruption: Current specializations (frontend, backend, etc.) are being disrupted. Now is the time to generalize and touch all aspects of development. Specializations will re-emerge, but not for several years as tools and methods settle.

• Invest in learning AI tooling now: If you’re not actively learning to work with agents and orchestrate AI workflows, you’re falling behind. This is comparable to when Google replaced going to Barnes & Noble for programming books, or when Rails dramatically accelerated web development productivity.

Career Advice:

• If you’re a developer who mostly fixes bugs and does grunt work, start thinking about product ideas and business domain improvements you can prototype with AI tools like Lovable. Showing initiative with a working prototype is your ticket to career growth.
• People who have survived previous industry shifts (web, mobile, cloud) are adapting faster because they recognize the pattern. Those with 10+ years on a single stack who never shifted are struggling the most.
• Newer developers are doing better than expected because they have less to unlearn.
• Writing, creative thinking, and domain expertise are becoming more valuable than pure coding skill.

Software Architecture Patterns:

• Use CLI-first design when building AI-assisted tools — Rob had AI build an ETL solution as a CLI first, then broadened it into a full application.
• AI brainstorming workflow: use AI for inspiration, then resonate on ideas together, then brainstorm, then build a plan — rather than trying to think of everything yourself.

Chapter Summaries

1998 Retrospective (Opening)

Carl and Richard review events from 1998, including the Good Friday Agreement, Google’s founding, the DMCA, Windows 98, Visual Basic 6, the iMac launch, the beginning of the International Space Station, and the first MP3 player.

Better Know a Framework: Tool3 (Tixle)

Richard highlights Tool3/Tixle, an open-source C# real-time animation toolkit for creating motion graphics, audio-reactive VJ content, and procedural animations.

The Human Side of AI Adoption

Rob explains that AI consulting is primarily an organizational and human challenge. He focuses on finding quick wins with painful tasks (like a 6-week ETL reduced to 90 minutes) to build buy-in and convert skeptics.

Prompting Philosophy

Rob shares his key prompting advice: tell the AI who it is, what you need, then always ask “What am I missing?” and “What do you recommend?” He uses AI as an architecture sparring partner for planning and decision-making.

Addressing Layoff Fears and Team Disruption

Discussion of Paulo Pinto’s comment about AI-driven layoffs. Rob argues most layoffs are driven by post-2020 overhiring, not AI, and that companies should use productivity gains to tackle backlogs rather than cut staff. Subject matter experts are already building their own apps, signaling a need for organizational restructuring.

AI, Creativity, and the Human Experience

Carl and Rob discuss the boundary between AI and human creativity in music and writing. Rob shares a story about creating a country song with AI for his girlfriend, exploring how people value the human intention behind creation even when tools are AI-powered.

Career Advice for Developers

The hosts advise developers to think about business problems, prototype solutions, and invest in learning AI tooling. Everyone becomes a generalist during disruption, and domain expertise plus creative thinking are the most valuable skills right now.