How Stripe built “minions”—AI coding agents that ship 1,300 PRs weekly from Slack reactions | Steve Kaliski (Stripe engineer)

Brought to you by:

• Optimizely—Your AI agent orchestration platform for marketing and digital teams

• Rippling—Stop wasting time on admin tasks, build your startup faster

Steve Kaliski has spent over six years building developer infrastructure at Stripe. In this conversation with Claire, he breaks down Stripe’s “minions”: AI coding agents that ship about 1,300 pull requests per week, often kicked off with nothing more than a Slack emoji. He explains why the real bottleneck in engineering isn’t coding, how cloud development environments unlock parallel AI workflows, and what it takes to safely review thousands of AI-generated PRs. He also demos AI agents that can spend money, coordinate services, and complete tasks end-to-end without human involvement.

Biggest takeaways:

1. What’s good for human developers is good for AI agents (and vice versa). Stripe’s years of investment in developer experience—comprehensive documentation, blessed paths for common tasks, robust CI/CD, excellent tooling—directly translates to higher AI agent success rates. When you have clear docs on “how to add a new API field,” the agent can follow those same instructions. This creates a virtuous cycle: investments in DX improve agent performance, and investments in agent infrastructure (like cloud environments) benefit human developers too.

2. Activation energy is the real bottleneck, not coding speed. Steve hasn’t started work in a text editor in months. Instead, work begins in Slack threads, Google Docs, or support tickets—the natural places where ideas emerge. By allowing engineers to kick off development with a single emoji reaction, Stripe lowered the friction between “good idea” and “code in production.” This is especially powerful in large organizations, where coordination costs typically kill momentum before coding even begins.

3. Cloud development environments are non-negotiable for multi-threaded AI work. Running multiple AI agents in parallel requires cloud-based dev environments that can spin up in seconds, run isolated workloads, and never fall asleep. This infrastructure investment—which Stripe’s developer productivity team built long before AI agents—now enables engineers to run dozens of agents simultaneously without melting their MacBook Pros.

4. 1,300 AI-written PRs per week requires shifting review capacity, not eliminating it. Stripe still reviews every AI-generated PR, but the review process relies heavily on automated confidence signals: comprehensive test coverage, synthetic end-to-end tests, and blue-green deployments that enable quick rollbacks. The bottleneck shifts from writing code to reviewing it—and eventually to generating enough good ideas in the first place.

5. Machine-to-machine payments unlock ephemeral, API-first businesses. In Steve’s birthday party demo, Claude Code autonomously paid Browser Base, Parallel AI, and Postal Form for single-use services—no human signup, no subscription, no dashboard. Businesses can now optimize for agent consumers rather than human users, focusing on “hyper-useful single APIs” instead of landing pages and admin panels. The economics become transparent: tokens and dollars sit side by side, making the true cost of AI work visible.

6. Treat AI agents like new employees, with progressive trust. Start with limited access, expand permissions as the agent proves reliable, and maintain clear boundaries. Each minion runs in an isolated environment with specific data access—the finance agent can read bank statements but can’t send messages; the scheduling agent can text but has no financial data. This physical partitioning prevents accidental data leakage and creates accountability.

7. The future of software is disposable and hyper-personalized. Steve builds custom iOS apps for his toddler—music players limited to six specific songs—despite having no iOS development experience. He describes this as “the disposability of software”: when AI can build apps in hours, you can create single-purpose tools for incredibly specific use cases and throw them away when they’re no longer needed.

Detailed workflow walkthroughs from this episode:

• How Stripe’s AI ‘Minions’ Ship 1,300 PRs Weekly from a Slack Emoji: https://www.chatprd.ai/how-i-ai/stripes-ai-minions-ship-1300-prs-weekly-from-a-slack-emoji

• How to Build an Autonomous AI Agent That Pays for Services to Complete Tasks: https://www.chatprd.ai/how-i-ai/workflows/how-to-build-an-autonomous-ai-agent-that-pays-for-services-to-complete-tasks

• How to Automate Code Generation from a Slack Message into a Pull Request: https://www.chatprd.ai/how-i-ai/workflows/how-to-automate-code-generation-from-a-slack-message-into-a-pull-request

How to turn Claude Code into your personal life operating system | Hilary Gridley

Brought to you by:

WorkOS—Make your app enterprise-ready today

Lovable—Build apps by simply chatting with AI

Hilary Gridley returns to the podcast to share how her approach to productivity has completely evolved since her last appearance. Now a new mom and entrepreneur, she walks Claire through how she uses Claude Code as a personal operating system, managing everything from daily planning to life admin without complex tools or rigid workflows. Instead of building elaborate systems, Hilary leans into what she calls the “anti-system system”: letting AI observe her behavior, learn her preferences over time, and gradually take work off her plate. Together, Claire and Hilary explore how simple inputs—like capturing tasks with a shortcut or “yapping” to Claude throughout the day—can replace traditional productivity stacks and integrations.

Biggest takeaways:

1. The 10x impact framework: For any task, ask “If I were 10 times better at it, would it have 10 times the impact?” If no, automate it. If yes, keep it human. This applies to both work tasks and life tasks—including whether baking bread will bring you joy or feel like a chore.

2. Complexity has to earn its keep. Hilary only connects APIs and builds complex integrations after testing the “janky version” of a workflow for a week. Her hit rate is only 20% on workflows she actually continues using, so starting simple saves massive time.

3. The yappers API beats OAuth every time. Instead of connecting all your tools in the background, just talk to Claude about what you’re doing throughout the day. Hilary keeps Claude Code open in her terminal and narrates her work, letting Claude observe and take notes without complex integrations.

4. Let AI learn your preferences by observing, not by your defining them. Hilary never sat down to write out her ideal schedule. Claude just watches what she actually does (not what she says she’ll do) and adjusts preferences automatically. Real behavior beats aspirational planning.

5. Calendar management is the ultimate to-do list. You can’t say you take something seriously if you’re not putting time into it. But manually adding tasks to your calendar is tedious—so let Claude do it automatically based on what you say you want to accomplish.

6. Screenshots are your friend for getting started. Don’t wait for API access or permissions at work. Build a janky version with screenshots and voice dictation, prove it’s valuable, and then get the permissions you need. Half-baked ideas don’t deserve full access.

7. You don’t need coding knowledge to build Claude Code skills. Hilary just describes problems to Claude: “I keep forgetting to return things on time.” Claude asks a few questions, then builds the entire workflow—including checking return policies and drop-off locations automatically.

8. Test everything before integrating it into working systems. Hilary refuses to add new workflows to her daily routine until she’s tested them separately for a week. If something breaks, you don’t want it taking down systems that were already working.

9. Build the muscle memory by doing one thing with AI every day. The biggest barrier isn’t technical knowledge—it’s rewiring your brain to think “the alien in my computer could help with this.” Hilary went from “I’ll never use the terminal” to running her life in Claude Code in about a week.

Detailed workflow walkthroughs from this episode:

• How I AI: Hilary Gridley’s “Anti-System” for Automating Life with Claude Code: https://www.chatprd.ai/how-i-ai/gridleys-anti-system-for-automating-life-with-claude-code

• Create a Privacy-Protecting Demo Mode for Your Personal AI: https://www.chatprd.ai/how-i-ai/workflows/create-a-privacy-protecting-demo-mode-for-your-personal-ai

• Build Custom AI Automations by Having a Conversation with Claude: https://www.chatprd.ai/how-i-ai/workflows/build-custom-ai-automations-by-having-a-conversation-with-claude

• Automate Your Daily Planning with Claude Code and an iPhone Shortcut: https://www.chatprd.ai/how-i-ai/workflows/automate-your-daily-planning-with-claude-code-and-an-iphone-shortcut

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

Hilary Gridley is an entrepreneur, former product leader, and new mom who previously appeared on the podcast discussing AI for managers. She returns to share how she's transformed her approach to personal productivity using Claude Code as her primary tool for managing both professional work and life admin. Hilary demonstrates her "anti-system system"—a philosophy that prioritizes simplicity over complex setup, allowing AI to learn preferences through observation rather than upfront configuration.

What you’ll learn:

1. How to capture to-dos instantly using a simple iPhone back-tap shortcut that requires zero app switching

2. The “10x impact framework” for deciding what tasks to automate versus where to invest your human effort

3. How to use Claude Code’s observation capabilities to build a preference file that improves over time without manual setup

4. Why the “yappers API” (talking about what you’re doing while working) eliminates the need for complex OAuth integrations

5. A workflow for breaking down overwhelming tasks into 10-minute first steps that actually get completed

6. How to create Claude Skills by simply describing problems rather than writing code or following tutorials

7. Techniques for using “recording mode” to demo workflows without exposing personal information

Brought to you by:

WorkOS—Make your app Enterprise Ready today

Lovable—Build apps by simply chatting with AI

In this episode, we cover:

(00:00) Introduction to Hilary Gridley

(02:43) The opportunity cost of time as a new mom and entrepreneur

(07:11) Philosophy of the anti-system system

(08:05) Demo: Planning your day with Claude Code

(10:00) Setting up simple iPhone shortcuts for task capture

(11:48) How Claude organizes reminders and learns preferences automatically

(16:19) Breaking down overwhelming tasks into manageable first steps

(23:40) The yappers API: talking to Claude instead of building integrations

(25:28) Daily logging and observation patterns

(27:45) Quick summary

(30:50) The power of screenshots

(32:55) 10x impact framework for automation decisions

(37:51) Applying the framework to different career stages

(39:29) Building a “recording on” skill for anonymizing demos

(44:11) Building a returns tracking skill from scratch

(48:31) Building the muscle memory to reach for AI tools

(50:18) Where to find Hilary

Tools referenced:

• Claude Code: https://claude.ai/code

• Obsidian: https://obsidian.md/

• iPhone Shortcuts: https://support.apple.com/guide/shortcuts/welcome/ios

• Cursor: https://cursor.sh/

Other references:

• Figma file Hilary demo’ed: https://www.writerbuilder.com/howiai

Where to find Hilary Gridley:

Substack: https://hills.substack.com/

Website: https://writerbuilder.com

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

Steve Kaliski is a software engineer at Stripe who has spent the past six and a half years building developer tools and payment infrastructure. He’s part of the team that created “minions”—Stripe’s internal AI coding agents, which now ship approximately 1,300 pull requests per week with minimal human intervention beyond code review. In this episode, Steve demonstrates how Stripe engineers activate development work from Slack and leverage cloud-based development environments for parallel agent workflows, and demos machine-to-machine payments where AI agents transact autonomously with third-party services.

What you’ll learn:

1. How Stripe’s “minions” write 1,300 pull requests per week with minimal human intervention

2. Why a good developer experience for humans creates better outcomes for AI agents

3. The critical role of cloud development environments in unlocking AI-powered engineering velocity

4. The machine payment protocol that lets AI agents spend money to accomplish tasks

5. The code review strategy for handling thousands of agent-written PRs

6. Why non-engineers at Stripe are starting to use minions to ship code

7. The future of software businesses built primarily for agent consumers

Brought to you by:

Optimizely—Your AI agent orchestration platform for marketing and digital teams

Rippling—Stop wasting time on admin tasks, build your startup faster

In this episode, we cover:

(00:00) Introduction to Steve

(02:39) Stripe’s minions and their effect on Stripe as a whole

(04:42) Why activation energy matters more than execution

(05:44) What is a minion? The technical architecture

(06:52) Demo: Activating a minion from Slack with an emoji

(09:04) Why good developer experience benefits both humans and agents

(11:22) Walking through the agent loop and system prompts

(13:42) Why Stripe chose Goose as their agent harness

(16:00) The role of Stripe’s developer productivity team

(17:15) Why cloud environments unlock multi-threaded AI engineering

(21:14) One-shot prompting: from Slack to shipped PR

(22:04) How Stripe handles code review for 1,300 AI-written PRs weekly

(23:44) Non-engineers using minions across the company

(24:53) Demo: Planning a birthday party with Claude and machine payments

(32:15) Quick recap

(35:08) The future of ephemeral, API-first businesses for agents

(36:36) Lightning round and final thoughts

Tools referenced:

• Goose (AI agent harness): https://github.com/block/goose

• Claude Code: https://claude.ai/code

• Cursor: https://cursor.sh/

• VS Code: https://code.visualstudio.com/

• Slack: https://slack.com/

• Browserbase: https://browserbase.com/

• Parallel AI: https://www.parallel.ai/

• PostalForm: https://postalform.com/

• Stripe Climate: https://stripe.com/climate

Other references:

• Stripe machine payments: https://docs.stripe.com/payments/machine

• Blue-Green Deployment: https://martinfowler.com/bliki/BlueGreenDeployment.html

• Git worktrees: https://git-scm.com/docs/git-worktree

Where to find Steve Kaliski:

Twitter: https://twitter.com/stevekaliski

LinkedIn: https://www.linkedin.com/in/steve-kaliski-079a7710/

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

How Microsoft’s AI VP automates everything with Warp | Marco Casalaina

Brought to you by:

• Rovo—AI that knows your business

• Lovable—Build apps by simply chatting with AI

Marco Casalaina, VP of Core AI Products at Microsoft, walks through how he uses Warp, Microsoft 365 Copilot, and ChatGPT to spin up lightweight “micro-agents” on the fly that handle everything from Azure admin to document scanning to video compression.

Biggest takeaways:

1. Automation frees you to focus on higher-value activities. As Marco puts it: “I do something else while these agents are doing whatever it is I need them to do.” When scanning his daughter’s homework, he was able to help her with math problems instead of fighting with scanner software.

2. Hardware automation is an underappreciated AI use case. Marco demonstrates how Warp can control his scanner through NAPS2, a CLI tool for scanners. This allows him to scan documents, flip pages, and combine PDFs without ever touching the scanner software.

3. File manipulation is a powerful AI capability. When Marco needed to compress a 1.7GB video file, he simply asked Warp to analyze why it was so large and re-encode it with FFmpeg. The AI identified the issue (excessive bitrate and resolution) and compressed it to 13MB while maintaining quality.

4. Ad hoc agents are changing how we approach automation. Marco defines these as “unnamed agents created on the fly to do something for me.” Rather than building complex, permanent workflows, these micro-agents handle specific tasks when needed and then disappear. This approach is becoming a trend across general-purpose AI tools, blurring the line between consuming and building agents.

5. CLI tools with AI create better interfaces than traditional GUIs. Marco points out that designing graphical interfaces for complex systems like Azure permissions is extremely difficult. AI with access to command-line tools can abstract away these design challenges.

Detailed workflow walkthroughs from this episode:

• How Microsoft’s AI VP Automates Everything with 5 Micro-Agent Workflows: https://www.chatprd.ai/how-i-ai/microsofts-ai-vp-automates-everything-with-5-micro-agent-workflows

• How to Create an Automated Meeting Scheduler with Microsoft 365 Copilot: https://www.chatprd.ai/how-i-ai/workflows/how-to-create-an-automated-meeting-scheduler-with-microsoft-365-copilot

• How to Scan and Merge Two-Sided Documents into a Single PDF with AI: https://www.chatprd.ai/how-i-ai/workflows/how-to-scan-and-merge-two-sided-documents-into-a-single-pdf-with-ai

• How to Automate Azure User Role Management with AI in the Terminal: https://www.chatprd.ai/how-i-ai/workflows/how-to-automate-azure-user-role-management-with-ai-in-the-terminal

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

Marco Casalaina, VP of Core AI Products and AI Futurist at Microsoft, demonstrates how he uses AI tools to automate administrative tasks that typically consume valuable time. Rather than using Warp as a coding assistant (its primary marketed purpose), Marco leverages it to manage Azure resources, scan documents, compress videos, and more. He shows how these “micro-agents” can reduce friction in everyday workflows, allowing him to focus on higher-value activities. Marco also demonstrates how Microsoft 365 Copilot and ChatGPT can create triggered workflows that respond to emails or check for information on a schedule, highlighting how the line between consuming and building AI agents is blurring.

What you’ll learn:

1. How to use Warp to manage Azure resources and assign permissions without navigating complex web interfaces

2. Techniques for automating document scanning and processing directly from the terminal

3. Methods for analyzing and compressing video files using AI-generated FFmpeg commands

4. How to create simple rules that dramatically improve AI performance for specialized tasks

5. Ways to build triggered workflows in Microsoft 365 Copilot that automatically respond to emails

6. How to configure ChatGPT to perform scheduled tasks like checking for new content

7. Strategies for creating consistent AI interactions using AutoHotkey shortcuts

Brought to you by:

Rovo—AI that knows your business

Lovable—Build apps by simply chatting with AI

In this episode, we cover:

(00:00) Introduction to Marco Casalaina

(02:14) Why Marco chose Warp for administrative tasks

(03:57) Demo: Using Warp to manage Azure resources and permissions

(06:00) How CLI tools eliminate GUI friction for complex tasks

(07:18) Creating rules to improve AI performance for specialized tasks

(10:28) Demo: Document scanning automation

(13:00) Combining odd and even pages using a Python automation

(15:04) The value of ephemeral AI solutions vs. permanent tools

(17:12) Video compression using FFmpeg commands

(20:22) The concept of “ad hoc agents” for specific tasks

(22:31) Demo: Creating triggered workflows in Microsoft 365 Copilot

(25:51) Demo: Setting up scheduled tasks in ChatGPT

(27:17) How AI automation changes time management

(29:14) Teaching AI skills to the next generation

(30:30) Strategies for improving AI performance with AutoHotkey

Tools referenced:

• Warp: https://www.warp.dev/

• Microsoft Azure: https://azure.microsoft.com/en-us

• Azure CLI: https://learn.microsoft.com/en-us/cli/azure/

• Microsoft 365 Copilot: https://www.microsoft.com/en-us/microsoft-365/copilot

• ChatGPT: https://chat.openai.com/

Other references:

• NAPS2: https://www.naps2.com/

• PyPDF2: https://pypdf2.readthedocs.io/

• FFmpeg: https://ffmpeg.org/

Where to find Marco Casalaina:

LinkedIn: https://www.linkedin.com/in/marcocasalaina/

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

From Figma to Claude Code and back

Brought to you by Optimizely—Your AI agent orchestration platform for marketing and digital teams

Gui Seiz (designer) and Alex Kern (engineer) from Figma show how to pull a live interface from production, staging, or localhost into Figma, turn it into editable design frames, explore variations collaboratively, and push changes back into code using Claude Code and MCPs—creating a continuous design ↔ code loop.

Biggest takeaways:

1. The design handoff is dead—replaced by continuous sync. Instead of designers creating comprehensive design packages with every state documented, AI enables bidirectional flow between Figma and code. Pull production code into Figma to see what actually exists, make changes in Figma, then push those changes directly back to code. No more version-final-final-v3.

2. Direct manipulation still beats prompting for precision. While AI can generate designs from prompts, dragging elements around in Figma remains the gold standard for fine-tuning. As Gui notes, “No one wants to prompt for the exact hex code or shade of yellow”—it’s just easier to use the color picker and eyeball it.

3. Use Figma’s MCP to keep design files current with production. The biggest problem in design-code workflows is drift—production gets ahead of Figma, or Figma contains dreams that never shipped. With MCP, you can programmatically pull any production state into Figma, ensuring that designers always work from what actually exists.

4. Turn your engineering wiki into executable skills. Every team has that onboarding page: “This is what you do before pushing a PR.” Alex built a /ship skill that automatically runs pre-flight checks, pushes to Git, monitors CI, and even fixes minor lint issues—up to five times with a one-hour timeout. Take every SOP and make it a skill.

5. Structure your codebase for AI assistance. Alex spends 20% to 30% of his time optimizing code structure so AI can accomplish more with less. This isn’t about writing better code for humans; it’s about making your codebase more legible to AI agents so each prompt delivers better results.

Detailed workflow walkthroughs from this episode:

1. How Figma’s Team Syncs Design and Code with Claude Code and Codex: https://www.chatprd.ai/how-i-ai/how-figma-team-syncs-design-and-code-with-claude-code-and-codex

2. Automate Your Pre-Merge PR Checklist with a Custom AI `/ship` Skill: https://www.chatprd.ai/how-i-ai/workflows/automate-your-pre-merge-pr-checklist-with-a-custom-ai-ship-skill

3. Automate Design Documentation by Exporting All Component States from Code to Figma: https://www.chatprd.ai/how-i-ai/workflows/automate-design-documentation-by-exporting-all-component-states-from-code-to-figma

4. Create a Bidirectional Sync Between Production Code and Figma Designs with AI: https://www.chatprd.ai/how-i-ai/workflows/create-a-bidirectional-sync-between-production-code-and-figma-designs-with-ai

From journalist to iOS developer: How LinkedIn’s editor builds with Claude Code

Brought to you by:

• WorkOS—Make your app enterprise-ready today

• Vanta—Automate compliance and simplify security

Daniel Roth (Editor in Chief and VP of Content Development at LinkedIn) shares how he builds and ships iOS apps to the App Store without writing code. He walks through the workflow he uses with Claude Code—including a dual-agent system where one AI writes code and another reviews it—along with how he plans features, manages development with branches, and turns ideas into working apps.

Biggest takeaways:

1. Create dueling AI agents to build better code. Daniel uses “Bob the Builder” to generate code and “Ray the Reviewer” to critique it for security and architecture issues. This two-agent system creates checks and balances similar to how engineering teams work, with Bob focusing on implementation and Ray ensuring quality. The friction between copying plans between agents also helps Daniel learn more about the code being generated.

2. Use AI to prevent dropping the ball at work. Daniel’s most valuable AI workflow isn’t for coding—it’s for managing his responsibilities as a leader of 400 people. He ends each day by asking Copilot, “What did I drop the ball on?” The AI scans his emails, Teams messages, and documents to identify unanswered messages and pending tasks. This 30-minute evening routine helps him wrap up loose ends before leaving work.

3. Build personalized apps that solve your own problems first. Daniel created “Commutely” to solve his specific problem of knowing whether to run for the New York subway. As he explains, “It was a perfect product-market fit because I was the entire product.”

4. Keep a running feature tracker with AI-powered prioritization. Daniel maintains a Claude chat that tracks all feature ideas with estimated build time and potential impact. His prompt instructs Claude to “keep track of ideas and offer guidance: time estimate to build, estimated back-and-forth hours, potential impact score on 1–3 scales for customer happiness and growth impact.” This creates a prioritized backlog he can pull from whenever he has time to build.

5. Document everything in Markdown files. Daniel saves all AI conversations as Markdown files, explaining, “Every time I’m working with Claude, I’m saying, ‘Write it into a file. Log everything.’” This solves two problems: Claude’s limited context window and his own memory limitations when returning to projects after breaks. This documentation habit creates a knowledge repository he can reference later.

Detailed workflow walkthroughs from this episode:

1. How I AI: Daniel Roth’s Dueling Agent Workflow for Building iOS Apps: https://www.chatprd.ai/how-i-ai/daniel-roth-dueling-agent-workflow-for-building-ios-apps

2. Build iOS Apps with a Dueling AI Agent Workflow: https://www.chatprd.ai/how-i-ai/workflows/build-ios-apps-with-a-dueling-ai-agent-workflow

3. How to Use Claude for AI-Powered Feature Prioritization: https://www.chatprd.ai/how-i-ai/workflows/how-to-use-claude-for-ai-powered-feature-prioritization

4. How to Use a Simple Copilot Prompt to Never Drop the Ball Again: https://www.chatprd.ai/how-i-ai/workflows/how-to-use-a-simple-copilot-prompt-to-never-drop-the-ball-again

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

Daniel Roth, editor in chief at LinkedIn, went from business writer to iOS app developer, without ever learning how to code. Using Claude Code, Daniel built and shipped multiple production-ready iOS apps to the App Store, including Commutely, a personalized train-tracking app for New York commuters.

What you’ll learn:

1. How to set up a dual-agent Claude Code system (builder + reviewer)

2. Why being a “picky customer” is the right mindset for non-technical builders

3. How Daniel prioritizes features using AI-ranked impact vs. build time

4. Why saving everything as Markdown files creates long-term context

5. The importance of branch-based development—even when AI writes the code

6. How Daniel ships to the App Store without formal engineering experience

7. His end-of-day “What did I drop the ball on?” Copilot workflow

Brought to you by:

WorkOS—Make your app enterprise-ready today

Vanta—Automate compliance and simplify security

In this episode, we cover:

(00:00) Introduction to Daniel Roth

(02:46) Daniel’s AI development workflow overview

(05:56) Using Claude to prioritize feature ideas

(08:58) Building vs. marketing

(09:47) Creating a retention plan for his app

(10:38) Introducing Bob the Builder and Ray the Reviewer

(13:50) How Bob and Ray work together to build features

(14:37) Why Daniel focuses on learning the process

(16:34) The importance of using branches for development

(17:39) Managing AI agents like managing a team

(21:12) Navigating the App Store

(23:06) Being a “picky customer” rather than a PM

(25:00) Testing in Xcode and shipping to the App Store

(28:14) Quick recap

(30:00) Creating terminal aliases with Claude

(31:38) Demo of his Commutely app

(32:10) Using Copilot to manage work responsibilities

(35:05) How Daniel talks to AI without personifying it

Tools referenced:

• Claude: https://claude.ai/

• Claude Code: https://claude.ai/code

• Cursor: https://cursor.sh/

• Xcode: https://developer.apple.com/xcode/

• Canva: https://www.canva.com/

• Microsoft Copilot: https://copilot.microsoft.com/

• Terminal: https://support.apple.com/guide/terminal/welcome/mac

• Obsidian: https://obsidian.md/

Other reference:

• Commutely (iOS app): https://apps.apple.com/us/app/commutely/id6755789873

Where to find Daniel Roth:

LinkedIn: https://www.linkedin.com/in/danielroth1/

Newsletter: https://www.linkedin.com/newsletters/forward-deployed-editor-7378272989982683137/

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

Most teams are still passing static design files back and forth, and most Figma files are already out of date by the time they reach engineering. Gui Seiz (designer) and Alex Kern (engineer) from Figma walk through the exact workflow their team uses to bridge that gap with AI, live onscreen. They demo how to pull a running web app directly into Figma using the Figma MCP, edit it collaboratively, and push it back to code. The old linear waterfall workflow is gone. What replaces it is a fluid, bidirectional loop where design and code inform each other in real time.

What you’ll learn:

1. How to use Figma’s MCP to pull production code directly into Figma files

2. A workflow for pushing design changes from Figma back into your codebase using Claude Code without manual CSS adjustments

3. How to export multiple code states (like all five states of a signup flow) into Figma so designers can work with what actually exists in production

4. Why AI has shifted design work upstream to planning and downstream to craft, eliminating the rushed middle phase of execution

5. How to create custom skills that automate pre-flight checks, lint fixes, and CI monitoring before pushing code to production

6. How to structure your codebase so AI can write 90% of your code more effectively

Brought to you by:

Optimizely—Your AI agent orchestration platform for marketing and digital teams

In this episode, we cover:

(00:00) Introduction to Gui and Alex from Figma

(02:56) How AI has transformed Figma’s internal workflows

(05:17) The collapse of linear design-to-code workflows

(07:28) Demo: Pulling production code into Figma using MCPs

(10:49) Using Figma for precise design manipulation and team collaboration

(14:10) Demo: Pushing Figma designs back into code with Claude Code

(16:06) How AI has changed the role of software engineers

(18:43) The shift to upstream planning and downstream craft

(22:31) Demo: Exporting multiple code states back into Figma

(25:23) Synchronous vs. asynchronous collaboration with AI

(28:00) Eliminating design and engineering toil with AI

(29:03) Demo: Custom skills for automating pre-flight checks

(34:06) Code first or design first?

(35:24) Using AI to learn and explore codebases

Tools referenced:

• Figma: https://www.figma.com/

• From Claude Code to Figma: Turning production code into editable Figma designs: https://www.figma.com/blog/introducing-claude-code-to-figma/

• Codex: https://codex.ai/

• Claude Code: https://claude.ai/code

• Buildkite: https://buildkite.com/

Other references:

• Balsamiq: https://balsamiq.com/

Where to find Gui Seiz:

X: https://x.com/guiseiz

LinkedIn: https://www.linkedin.com/in/guiseiz/

Where to find Alex Kern:

X: https://x.com/kernio

LinkedIn: https://www.linkedin.com/in/alexanderskern/

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

Every Monday, host Claire Vo shares a 30- to 45-minute episode with a new guest demoing a practical, impactful way they’ve learned to use AI in their work or life. No pontificating—just specific and actionable advice.

Mastering Midjourney: How to create consistent, beautiful brand imagery without complex prompts

Brought to you by:

• Vanta—Automate compliance and simplify security

• Lovable—Build apps by simply chatting with AI

Jamey Gannon is an AI creative director who has spent thousands of hours working inside Midjourney. In this episode, she walks through the system she uses to generate consistent, high-quality brand imagery using mood boards, style references, and personalization codes. She and Claire cover her full workflow for turning loose inspiration into a scalable visual system, how to avoid the biggest mistakes people make with AI image tools, and why a single image reference can often outperform a paragraph of prompting.

Detailed workflow walkthroughs from this episode:

• How I AI: Jamey Gannon’s Workflow for Consistent Brand Imagery in Midjourney: https://www.chatprd.ai/how-i-ai/consistent-brand-imagery-in-midjourney

• How to Generate Realistic AI Self-Portraits for Content: https://www.chatprd.ai/how-i-ai/workflows/how-to-generate-realistic-ai-self-portraits-for-content

• How to Fix and Refine AI-Generated Images: https://www.chatprd.ai/how-i-ai/workflows/how-to-fix-and-refine-ai-generated-images

• How to Create a Consistent Brand Aesthetic in Midjourney: https://www.chatprd.ai/how-i-ai/workflows/how-to-create-a-consistent-brand-aesthetic-in-midjourney

Biggest takeaways:

1. Avoid prompting. Jamey’s entire workflow is designed to minimize text prompting: “I try to avoid prompting at all costs in my process.” Instead of writing complex JSON prompts or detailed descriptions, she relies on style references, image references, and personalization codes to communicate visually with AI.

2. Style references (SREFs) often outperform mood boards. While mood boards are a good starting point, using SREFs as the mood board can give much better results. This is because mood boards with varied styles can cause Midjourney to “average things out,” while SREFs provide more direct style guidance.

3. Personalization codes create your unique aesthetic. By carefully rating images during the personalization process, you can create codes that consistently deliver your preferred style. Jamey recommends “only picking things that I would like if I generated it” and being mindful of potential “style bleeding” from consistently selecting certain elements.

4. Editorial and camera references are prompting shortcuts. Instead of describing complex aesthetics, simply reference known publications or camera models: “When you just say ‘Dazed editorial,’ depending on how famous the publication is, Midjourney’s gonna know what you’re talking about.” This communicates lighting, composition, and style instantly.

5. When something’s not working, remove it completely. Rather than trying to fix problematic elements through prompting, Jamey takes a more direct approach: “If you’re getting too much green, just boot the thing that is so obvious and overwhelming.” This might mean removing an image from your references or cropping out distracting elements.

6. Use AI routing to get the best from each model. Jamey uses different tools for different purposes: Midjourney for initial creation, Nano Banana for editing specific elements, and Flora for packaging and delivering brand imagery to clients. Each tool has strengths that complement the others.

7. See things from the AI’s perspective. When troubleshooting, try to understand what the AI is “seeing” in your references: “I think where a lot of people get tripped up is trying to make AI work the way you want it to work instead of actually understanding how it works.” This perspective shift often leads to better results.

▶️ Listen now on YouTube | Spotify | Apple Podcasts

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

Jamey Gannon is an AI creative director who specializes in creating consistent, beautiful brand imagery using AI tools. In this episode, Jamey demonstrates her streamlined workflow for generating cohesive brand assets using Midjourney, Nano Banana, and other AI image tools. She walks through her process of creating mood boards, using style references, developing personalization codes, and strategically iterating to achieve a consistent aesthetic. Rather than relying on complex prompts, Jamey shows how visual references and strategic shortcuts can produce better results with less effort.

What you’ll learn:

1. How to create effective mood boards that communicate your desired aesthetic to AI image generation tools

2. Why style references (SREFs) often produce more consistent results than general mood boards in Midjourney

3. A systematic approach to testing and refining your visual style

4. How to use personalization codes in Midjourney to develop your own unique aesthetic preferences

5. Techniques for combining image references, style references, and minimal prompting to achieve consistent brand imagery

6. A workflow for using Nano Banana to fix specific elements in Midjourney-generated images without extensive editing

7. How to package and deliver your brand imagery system to clients so they can continue generating consistent assets

Brought to you by:

Vanta—Automate compliance and simplify security

Lovable—Build apps by simply chatting with AI

In this episode, we cover:

(00:00) Introduction to Jamey Gannon

(02:31) Creating mood boards as the foundation for AI image generation

(08:45) Using SREFs for better consistency

(11:15) Test prompts for evaluating style consistency

(12:33) The iterative process of creating and refining images

(24:28) Combining techniques for consistent brand imagery

(28:25) Scaling out your aesthetic across different subjects

(35:48) Using Nano Banana for targeted image refinements

(38:23) Creating realistic AI self-portraits for content

(43:04) Building a visual reference library for inspiration

(46:50) Troubleshooting techniques when AI isn’t cooperating

Tools referenced:

• Midjourney: https://www.midjourney.com/

• Nano Banana: https://gemini.google/overview/image-generation/

• Flora: https://flora.ai/

• Pinterest: https://www.pinterest.com/

• Cosmos: https://www.cosmos.so/

Other reference:

• Style references (SREFs) in Midjourney: https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference

Where to find Jamey Gannon:

Website: https://www.brand-sprints.com/links

LinkedIn: https://www.linkedin.com/in/jameygannon/

X: https://x.com/jameygannon

Instagram: https://www.instagram.com/jameygannon

Maven Course (get 10% off with this link): https://bit.ly/4b18RfM

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

Every Monday, host Claire Vo shares a 30- to 45-minute episode with a new guest demoing a practical, impactful way they’ve learned to use AI in their work or life. No pontificating—just specific and actionable advice.

5 OpenClaw agents run my home, finances, and code | Jesse Genet

Brought to you by: Optimizely—Your AI agent orchestration platform for marketing and digital teams

Jesse Genet is a homeschooling parent and entrepreneur who operates five specialized OpenClaw agents, each on its own Mac Mini, to manage homeschool curriculum, family finances, scheduling, development projects, and household operations. She treats each agent like a new hire: defined role, scoped access, decision log, and progressive trust. In this episode she shares how she photographs curriculum books to auto-generate lesson plans, builds custom apps with zero prior terminal experience, partitions sensitive data across machines, and bridges the digital and physical world by inventorying every toy and supply in her house.

Detailed workflow walkthroughs from this episode:

• How I AI: Jesse Genet’s 5 OpenClaw Agents for Homeschooling, App Building, and Physical Inventories: https://www.chatprd.ai/how-i-ai/jesse-genets-5-openclaw-agents-for-homeschooling-app-building-and-physical-inventories

• Automate Homeschool Lesson Planning and Material Creation with an AI Agent: https://www.chatprd.ai/how-i-ai/workflows/automate-homeschool-lesson-planning-and-material-creation-with-an-ai-agent

• Build a Custom ‘Slop-Free’ Kids’ TV App Without Coding Experience: https://www.chatprd.ai/how-i-ai/workflows/build-a-custom-slop-free-kids-tv-app-without-coding-experience

• Create an AI-Powered Inventory of Your Physical Items: https://www.chatprd.ai/how-i-ai/workflows/create-an-ai-powered-inventory-of-your-physical-items

Biggest takeaways:

1. Treat your AI agent like a new hire, not an extension of yourself. Jesse’s entire agent management philosophy comes from her experience hiring employees. She gives agents their own identities, separate data access, and communication channels—never full access to her email or accounts. Progressive trust is the model: start limited, expand as the agent proves reliable.

2. Physical partitioning is a real security strategy. Running each agent on its own Mac Mini sounds extreme, but it solves a real problem: preventing one agent from accidentally leaking sensitive data through another agent’s communication channel. The finance agent can read bank statements but can’t text anyone. The scheduling agent can text, but has no financial data. This is a practical framework anyone managing multiple agents should think through.

3. Photos are the most underrated input for AI agents. Jesse’s core workflow is shockingly simple: take a photo, send it to the agent, get structured output. She photographs lesson activities, book pages, physical supplies, and curriculum materials. The agent handles all the heavy lifting of logging, categorizing, and connecting that information. No typing, no structured input—just photos.

4. You don’t need to be technical to build real software with a coding agent. Jesse had never opened a terminal before six months ago. With Cole, her coding agent, she built a custom kids’ TV app, iterated over four days, and deployed it to a Google TV Streamer. Her approach: describe what you want, push back when the agent says something isn’t possible, and keep going.

5. Inventory your physical world so AI can reach into it. One of Jesse’s most powerful moves was photographing all her educational supplies and having Sylvie build an inventory. Now when she asks for a lesson plan, the agent can say, “Also, pull out the tracing board from the cupboard.” This bridges the gap between the digital agent and the physical world in a way that’s immediately useful.

6. Use “decision files” to prevent agents from relitigating settled questions. Jesse maintains a decisions file in Obsidian that each agent knows about. When she makes a final call, she flags it, and the agent knows not to revisit it. This is a simple, powerful pattern for anyone dealing with agents that keep second-guessing or re-asking about things you’ve already resolved.

▶️ Listen now on YouTube | Spotify | Apple Podcasts

How Coinbase scaled AI to 1,000+ engineers | Chintan Turakhia

Brought to you by:

• WorkOS—Make your app enterprise-ready today

• Rovo—AI that knows your business

Chintan Turakhia leads engineering at Coinbase, where he’s helped turn a 1,000-plus-person org into an AI-native team. In this episode, Chintan shares how Coinbase got 100 engineers to ship 75 PRs in 15 minutes, cut PR review time from 150 hours to 15, and built internal agents that turn user feedback into shipped features in minutes. Claire and Chintan dive into the “speed run” tactic that broke through skepticism, how to identify and replicate AI power users, and why the best engineering leaders are getting more hands-on with code in the AI era.

Detailed workflow walkthroughs from this episode:

• How I AI: Chintan Turakhia’s Playbook for AI Adoption at Coinbase: https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase

• Use ChatGPT to Become Your Own Personal Wine Sommelier: https://www.chatprd.ai/how-i-ai/workflows/use-chatgpt-to-become-your-own-personal-wine-sommelier

• Build an Automated User Feedback to Pull Request Pipeline: https://www.chatprd.ai/how-i-ai/workflows/build-an-automated-user-feedback-to-pull-request-pipeline

• Create a Data-Driven AI Adoption Playbook Using Cursor: https://www.chatprd.ai/how-i-ai/workflows/create-a-data-driven-ai-adoption-playbook-using-cursor

Biggest takeaways:

1. Organize “speed runs” to kickstart widespread AI adoption. Coinbase gathered 100 engineers for a 15-minute session where everyone used AI tools to push 75 PRs simultaneously. The high success rate of merged PRs proved AI’s effectiveness at scale and transformed the team’s mindset about what they could accomplish.

2. Show, don’t tell, when driving AI adoption in engineering teams. Chintan spent months personally using Cursor daily, discovering use cases and techniques before asking his team to adopt it. By tackling real bugs and showing concrete wins, he built credibility that made adoption natural rather than forced.

3. Focus AI adoption on eliminating “soul-sucking” work first. Rather than tackling complex engineering challenges, Chintan targeted the tedious tasks engineers hate: writing unit tests, fixing linting issues, and managing Git commands. Engineers naturally want to build better things faster, and removing friction from their workflow created immediate buy-in.

4. Create public channels to showcase AI wins. Coinbase established a “Cursor Wins” Slack channel where engineers shared their successes. This visibility created organic FOMO and peer learning as engineers saw colleagues shipping more code with less effort.

5. Measure the entire feedback-to-feature cycle time, not just lines of code. Coinbase reduced PR review times from 150 hours to 15 hours and built systems to capture user feedback and convert it to shipped features in minutes rather than weeks. This end-to-end acceleration creates a virtuous cycle where faster shipping leads to more user feedback and continuous improvement.

6. Engineering leaders should be writing more code, not less. Chintan’s calendar is now nearly empty, as AI has eliminated coordination overhead and meetings. Instead, he spends more time writing code, fixing bugs, and exploring technical approaches.

7. Use AI to analyze AI adoption patterns. Chintan used Cursor itself to analyze Cursor usage data, identifying natural user cohorts from inactive to super-users. This meta-application of AI allowed him to create targeted guidance for each segment, helping engineers progress from light usage to power usage. The analysis revealed that agent-heavy users were 16x more productive with AI than other users, providing concrete data to drive further adoption.

▶️ Listen now on YouTube | Spotify | Apple Podcasts

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

Chintan Turakhia is Senior Director of Engineering at Coinbase, where he’s led the transformation of a 1,000-plus-engineer organization to embrace AI tools at scale. When tasked with rewriting Coinbase’s self-custody wallet into a consumer social app in just six to nine months, Chintan turned to AI as a force multiplier. His team has achieved remarkable efficiency gains, including reducing PR review times from 150 hours to just 15 hours, and dramatically compressing the cycle from user feedback to shipped features.

What you’ll learn:

1. How to drive AI adoption in large, established engineering organizations

2. The “speed run” technique that got 100 engineers to push 70 PRs in 15 minutes

3. How to identify and replicate the behaviors of AI power users

4. Why engineering leaders must get hands-on with AI tools to drive adoption

5. How to build custom AI agents that integrate with your existing workflows

6. The metrics that actually matter when measuring AI’s impact on engineering velocity

7. How to compress the cycle from user feedback to shipped features

Brought to you by:

WorkOS—Make your app enterprise-ready today

Rovo—AI that knows your business

In this episode, we cover:

(00:00) Introduction to Chintan

(02:38) How Coinbase approached rewriting their app with AI assistance

(08:00) The importance of leadership conviction and hands-on demonstration

(10:30) The “PR speed run” technique that transformed team adoption

(17:57) Measuring success

(19:20) Demo: Real-time feedback-to-feature implementation

(23:14) Using Cursor to analyze AI adoption patterns

(33:15) Quick recap and appreciation

(36:00) Demo: Building a live feedback capture system using AI transcription

(40:50) Using custom Slack bots to automate engineering workflows

(47:10) Advice for driving AI adoption within your organization

(50:00) Personal use case: AI for wine selection based on taste preferences

(55:23) Lightning round and final thoughts

Tools referenced:

• Cursor: https://cursor.sh/

• Linear: https://linear.app/

• Slack: https://slack.com/

• ChatGPT: https://chat.openai.com/

• Claude: https://claude.ai/

• GitHub Copilot: https://github.com/features/copilot

Other references:

• Coinbase: https://www.coinbase.com/

• React Native: https://reactnative.dev/

• How custom GPTs can make you a better manager | Hilary Gridley (Head of Core Product at Whoop): https://www.lennysnewsletter.com/p/how-custom-gpts-can-make-you-a-better-manager

Where to find Chintan Turakhia:

LinkedIn: https://www.linkedin.com/in/chintanturakhia/

X: https://x.com/chintanturakhia

Base App (formerly Coinbase Wallet): https://base.app/

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

Jesse Genet is a homeschooling parent and entrepreneur who runs her household with five specialized OpenClaw agents. She layers them on top of her Obsidian “second brain,” deploys each on its own Mac Mini, and assigns every agent a distinct role—homeschool, finance, scheduling, development, and operations—so each one operates with clear scope and responsibility.

What you’ll learn:

1. How Jesse set up five OpenClaw agents, each with its own role, persona, SOUL.md file, and dedicated Mac Mini

2. The workflow for photographing an entire curriculum book and having an agent generate formatted, ready-to-teach lesson plans from the images

3. Using a coding agent to build a custom kids’ TV app from scratch and ship it to a real television in four days (with zero prior terminal experience)

4. Why Jesse treats agent onboarding like employee onboarding

5. The “decision file” trick and other incantations for managing agents that actually stick

6. Where multi-agent collaboration breaks down, and why no current messaging platform handles agent-to-agent handoffs well

7. How photographing every toy, book, and supply in the house lets the AI recommend real physical materials during lesson planning

8. The hands-free printing loop that took Jesse from scan → upload → email → print to “Sylvie, print this” in 30 seconds flat

Brought to you by:

Optimizely—Your AI agent orchestration platform for marketing and digital teams

In this episode, we cover:

(00:00) Meet Jesse and her “after Claw” life

(02:30) Layering OpenClaw on top of Obsidian

(04:44) Logging homeschool lessons automatically

(07:12) Turning books into a structured curriculum

(13:09) Using SOUL.md files to give each agent a personality

(14:39) Running multiple specialized AI agents

(16:43) Agent collaboration

(18:19) Partitioning data across Mac Minis

(27:00) Building a custom YouTube app with AI

(37:00) Creating a physical inventory from cupboard photos

(41:00) Printing from voice: reducing friction

(44:00) Managing agent memory and decision files

Tools referenced:

• OpenClaw: https://openclaw.ai/

• Obsidian: https://obsidian.md

• Slack: https://slack.com

• QuickBooks: https://quickbooks.intuit.com

• Google Gemini: https://gemini.google.com/

• Mac Mini: https://www.apple.com/mac-mini/

Other references:

• Claude Code for product managers: research, writing, context libraries, custom to-do system, and more | Teresa Torres: https://www.lennysnewsletter.com/p/claude-code-for-product-managers

Where to find Jesse Genet:

X: https://x.com/jessegenet

LinkedIn: https://www.linkedin.com/in/jessegenet/

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

“I haven’t written a single line of front-end code in 3 months”: How Notion’s design team uses Claude Code to design

Brought to you by:

• WorkOS—Make your app enterprise-ready today

• Orkes—The enterprise platform for reliable applications and agentic workflows

Brian Lovin, a product designer at Notion, built a shared AI-powered “prototype playground” that lets the entire design team turn Figma designs into working code using Claude Code. Instead of staying in static mockups, the team prototypes directly in a shared Next.js environment connected to real AI models—so they can test ideas in the browser, catch edge cases early, and design for what’s actually possible. In this episode, Brian breaks down how the system works, how he uses plan mode, slash commands, and custom Claude Skills to automate repetitive tasks, and why his core rule for working with AI is simple: when Claude asks you to do something, teach it to do that thing itself.

Detailed workflow walkthroughs from this episode:

• How Notion Designs with AI: Brian Lovin’s Prototype Playground and Claude Code Workflows: https://www.chatprd.ai/how-i-ai/how-notion-designs-with-ai-brian-lovins-prototype-playground-and-claude-code-workflows

• Automate Your Git and Deployment Workflow with a Custom AI Command: https://www.chatprd.ai/how-i-ai/workflows/automate-your-git-and-deployment-workflow-with-a-custom-ai-command

• Build an AI Workflow to Convert Figma Designs to Code with a Self-Correction Loop: https://www.chatprd.ai/how-i-ai/workflows/build-an-ai-workflow-to-convert-figma-designs-to-code-with-a-self-correction-loop

• Use Claude Code to Rapidly Build Interactive Prototypes from Ideas: https://www.chatprd.ai/how-i-ai/workflows/use-claude-code-to-rapidly-build-interactive-prototypes-from-ideas

Biggest takeaways:

1. Designs are shifting to code-first prototyping. While Brian still spends 60% to 70% of his time in Figma, he believes designers increasingly need to understand what AI models can actually do. This requires working with real models in code to “design something that’s plausible and possible.”

2. Encounter reality as early as possible in the design process. Brian’s philosophy is to move designs from “napkin sketches” toward production code as quickly as possible. When you try designs in a browser instead of Figma, you immediately notice problems with loading states, screen sizes, and interactions that static designs hide.

3. The “prototype playground” is a shared Next.js app that centralizes all design prototypes. Instead of designers working in isolated repositories with different setups, this shared environment makes it easy to discover what others are working on and reuse code. The repository organizes prototypes by designer name and provides shared components for Notion-style UI elements.

4. Brian found it impossible to design good AI experiences in Figma: “You can design what the chat input looks like ... but what you can’t design in Figma is what it actually will feel like to use that thing.” Code prototypes connected to real AI models are essential for understanding edge cases and failure modes.

5. When Claude asks you to do something, teach it to do that thing itself. Brian’s most important rule for working with AI is to avoid manual intervention. For example, instead of manually checking if a prototype works in the browser, teach Claude to launch Chrome, test the functionality, and verify the results.

6. Claude Skills can solve specific recurring problems. When AI consistently hallucinated icon names (using “search” instead of “magnifying glass”), Brian created a skill that programmatically searches for icons and their synonyms across thousands of files. This demonstrates how AI can be taught to overcome its own limitations.

7. Custom slash commands dramatically simplify complex workflows. Brian created commands like “/figma” that handle everything from checking if MCPs are installed to extracting designs, implementing them as code, and verifying the results through multiple iterations. This makes advanced AI techniques accessible to less-technical team members.

▶️ Listen now on YouTube | Spotify | Apple Podcasts

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

Brian Lovin is a designer at Notion AI who has transformed how the design team builds prototypes, by creating a shared code environment powered by Claude Code. Instead of designers working in isolated repositories or limited to static Figma designs, Brian built a collaborative “prototype playground” where the entire team can create, share, and iterate on functional prototypes. In this episode, Brian demonstrates how AI-assisted coding has dramatically accelerated the design process and why code-based prototyping is essential for building AI-powered products.

What you’ll learn:

1. How Brian built a shared Next.js app that serves as a collaborative prototyping environment for Notion’s design team

2. Why encountering “reality” early in the design process leads to better products

3. How to use Claude Code’s “plan mode” to get better results when prototyping

4. The power of custom Claude slash commands and skills to automate repetitive tasks

5. How to transform Figma designs into working code with a single prompt

6. Why AI-powered products can’t be effectively designed in static tools like Figma

7. Brian’s rule for working with AI: “When Claude asks you to do something, teach it to do that thing itself”

Brought to you by:

WorkOS—Make your app enterprise-ready today

Orkes—The enterprise platform for reliable applications and agentic workflows

In this episode, we cover:

(00:00) Introduction to Brian

(02:36) Building for B2B SaaS

(04:42) Notion’s prototype playground: what it is and how it works

(08:01) The technical background of designers using the playground

(10:52) Demo: building a podcast player prototype

(16:00) Actionable tips for better Claude Code results

(20:16) Analyzing the result

(20:30) Creating slash commands to simplify the workflow

(23:03) Turning Figma designs into production-ready code

(25:06) MCP frustrations and tips

(30:54) Demo: creating a custom “find icon” skill

(35:03) Demo: Creating a deploy command to simplify GitHub workflows

(41:09) Quick recap

(41:59) How code-based prototyping is changing design at Notion

(46:48) Brian’s tool preferences

(48:42) Prompting techniques when AI is not listening

Tools referenced:

• Claude Code: https://claude.ai/

• Cursor: https://cursor.sh/

• Next.js: https://nextjs.org/

• Figma: https://figma.com/

• Monologue: https://www.monologue.to/

• GitHub: https://github.com/

• GitHub Desktop: https://desktop.github.com/

• Tailwind CSS: https://tailwindcss.com/

• Bun: https://bun.sh/

Other references:

• Claude Skills explained: How to create reusable AI workflows: https://www.lennysnewsletter.com/p/claude-skills-explained

Where to find Brian Lovin:

Website: https://brianlovin.com/

LinkedIn: linkedin.com/in/brianlovin

X: https://twitter.com/brian_lovin

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

Claude Opus 4.6 vs GPT-5.3 Codex: Which is the better software engineer?

Brought to you by WorkOS—Make your app enterprise-ready today

Claire put GPT-5.3 Codex and Claude Opus 4.6 head-to-head on real work—redesigning her marketing site, refactoring complex components, and shipping production code. In five days, she shipped 44 PRs, touched 1,088 files, and added 93,000 lines of code (and deleted 87,000). Her big insight: It’s not Opus or Codex. It’s Opus for building and Codex for reviewing.

Detailed workflow walkthroughs from this episode:

• How I AI: GPT-5.3 Codex vs. Claude Opus 4.6—Shipping 44 PRs in 5 Days: https://www.chatprd.ai/how-i-ai/gpt-5-3-codex-vs-claude-opus-4-6

• How to Combine Claude Opus and GPT-5.3 Codex for High-Velocity Code Refactoring: https://www.chatprd.ai/how-i-ai/workflows/how-to-combine-claude-opus-and-gpt-5-3-codex-for-high-velocity-code-refactoring

• How to Redesign a Marketing Website Using Claude Opus 4.6 for Creative Development: https://www.chatprd.ai/how-i-ai/workflows/how-to-redesign-a-marketing-website-using-claude-opus-4-6-for-creative-development

Biggest takeaways:

1. AI coding tools have reached an inflection point for productivity. Claire shipped 44 PRs containing 98 commits across 1,088 files in just five days—adding 93,000 lines of code and removing 87,000. This included major features like MCP integrations and complete component refactors that would have taken months with a traditional team.

2. The perfect AI engineering workflow combines both models. Claire’s most productive approach was having Opus build features (getting them 80% to 90% complete) and then using Codex to review the code and find edge cases, before returning to Opus to implement the fixes. This mimics a junior–senior developer relationship.

3. Opus 4.6 is the “eager product engineer” who actually builds things. It’s excellent at planning and executing long-running, creative tasks. While its initial design output wasn’t great, it responded well to feedback and ultimately produced a much better redesign than Codex—one that Claire plans to ship to production.

4. GPT-5.3 Codex is the “principal engineer who won’t build anything.” Claire found that while Codex excels at reviewing code and finding edge cases, it struggles with creative, greenfield work. It follows instructions too literally and overfits to the last prompt, making it frustrating for tasks like redesigning a website from scratch.

5. The harness matters as much as the model. Claire found that Cursor’s interface worked better with both models than Codex’s native app. Features like plan mode, to-dos, and exploration tools in Cursor helped get better results from Opus than might have been possible in Claude Code.

6. Opus 4.6 Fast is blazingly fast but expensive. At roughly $150 per million output tokens (six times the price of standard Opus), it’s a significant investment. Claire embraces a “token abundance mindset” because the ROI is still massive compared with traditional development costs.

7. Codex brings Git concepts front and center in its interface. The app emphasizes repositories, branches, work trees, diffs, and pull requests—making it both powerful for experienced developers and educational for those learning Git concepts. This visual approach to Git is more accessible than command-line tools.

8. Skills and automations are becoming first-class citizens in AI coding tools. Codex presents these as visual components with icons and buttons rather than ZIP files, making them more accessible. The included recommended skills and automations provide good starting points for developers.

▶️ Listen now on YouTube | Spotify | Apple Podcasts

How this visually impaired engineer uses Claude Code to make his life more accessible | Joe McCormick

Brought to you by Tines—Start building intelligent workflows today

Joe McCormick is a principal engineer at Babylist who lost most of his central vision right before college. Instead of stepping away from software, he doubled down. And now, with tools like Claude Code, he’s building small, AI-powered Chrome extensions that make his life dramatically more accessible. In this episode, Joe live-builds “micro apps” that describe Slack images, fix typos instantly, and summarize links—all triggered with simple keyboard shortcuts.

Detailed workflow walkthroughs from this episode:

• How I AI: Building Custom AI Accessibility Tools for Slack with Joe McCormick & Claude Code: https://www.chatprd.ai/how-i-ai/custom-ai-accessibility-tools-for-slack-claude-code

• Build a Slack Link Summarizer from Scratch using Claude Code: https://www.chatprd.ai/how-i-ai/workflows/slack-link-summarizer-using-claude-code

• Create a Fast, Accessible AI Spell Checker for Any Website: https://www.chatprd.ai/how-i-ai/workflows/accessible-ai-spell-checker-for-any-website

• Build a Custom AI Tool to Describe Images in Slack: https://www.chatprd.ai/how-i-ai/workflows/ai-tool-to-describe-images-in-slack

Biggest takeaways:

1. AI is closing the gap between sighted and visually impaired engineers. As Joe explains, “The gap between a software engineer for a sighted person and a visually impaired person is closing day by day.” AI tools allow him to work more efficiently by eliminating accessibility barriers that previously required tedious workarounds.

2. Claude Code can be made more screen-reader-friendly with simple techniques. Joe demonstrates several accessibility improvements: using Control+G to edit prompts in a text editor instead of the terminal, creating a sound alert when Claude needs input, and leveraging consistent keyboard patterns (1 for yes, 2 for variations, 3 for no) to navigate options without visual cues.

3. Multimodal AI has transformed personal moments beyond work. Joe shares how Gemini allows him to read any book to his children, something previously impossible without memorizing the text. “I was always afraid of not being able to read stories to my kids. That ‘Sorry, I can’t’ has become ‘Sorry, I can,’ with the assistance of so many different tools now.’”

4. Keyboard shortcuts dramatically improve accessibility efficiency. Joe demonstrates how triggering extensions with keyboard combinations (Ctrl+Shift+D for image descriptions, Ctrl+Shift+S for spell-checking) eliminates multiple clicks and context switches. For someone using a screen reader and 10x magnification, these shortcuts transform tedious tasks into instant actions.

5. Running Slack in Chrome instead of as a desktop app enables powerful customizations. This clever workaround allows Joe to extend Slack’s functionality with Chrome extensions, something impossible with the desktop app. This approach could work for many other web-based tools that offer both desktop and browser versions.

6. Creating Claude Skills accelerates repetitive development tasks. After building his first two Chrome extensions, Joe created a Claude Skill to extract common patterns, making each subsequent extension faster to develop. This compounding efficiency means his fifth extension will take a fraction of the time of his first one.

▶️ Listen now on YouTube | Spotify | Apple Podcasts

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

Joe McCormick is a principal software engineer at Babylist who lost most of his central vision due to a rare genetic disorder right before starting college. He pivoted from mechanical engineering to computer science and now leads AI enablement at Babylist. Joe demonstrates how he uses AI to build micro Chrome extensions that make his everyday work and life more accessible, showing how personal software can address accessibility needs that mainstream products often overlook.

What you’ll learn:

1. How to build custom Chrome extensions in under 25 minutes using Claude Code

2. A practical workflow for creating AI-powered accessibility tools

3. How to use Claude Skills to accelerate repetitive development tasks

4. Techniques for making Claude Code more screen reader accessible

5. Why personal software is becoming increasingly viable with AI assistance

6. How multimodal AI is transforming accessibility for visually impaired users

Brought to you by:

Tines—Start building intelligent workflows today

In this episode, we cover:

(00:00) Introduction to Joe and his background

(02:34) Joe’s journey into computer science after vision loss

(04:50) The concept of personal software for accessibility

(06:09) Demo of image description Chrome extension for Slack

(10:40) Demo of AI-powered spell checker extension

(13:12) The efficiency of keyboard shortcuts for accessibility

(14:37) Live building a link summarization extension

(20:28) Using Claude Skills to extract common patterns

(25:30) Reviewing and modifying the development plan

(27:45) Removing cognitive friction for users through repeating patterns

(31:40) How to get fluent with AI tools

(34:55) Loading the extension into Chrome in developer mode

(36:19) Testing and debugging the extension

(40:44) Quick recap

(42:12) Lightning round and final thoughts

Tools referenced:

• Claude Code: https://claude.ai/code

• VS Code: https://code.visualstudio.com/

• Gemini: https://gemini.google.com/

• ChatGPT: https://chat.openai.com/

• Meta Ray-Ban Smart Glasses: https://www.meta.com/smart-glasses/

Other references:

• Chrome Extensions Documentation: https://developer.chrome.com/docs/extensions/

• ARIA (Accessible Rich Internet Applications): https://developer.mozilla.org/en-US/docs/Web/Accessibility/ARIA

• Windows Subsystem for Linux: https://learn.microsoft.com/en-us/windows/wsl/

• Screen Readers: https://www.afb.org/blindness-and-low-vision/using-technology/assistive-technology-products/screen-readers

• Claude Skills explained: How to create reusable AI workflows:https://www.lennysnewsletter.com/p/claude-skills-explained

Where to find Joe McCormick:

LinkedIn: https://www.linkedin.com/in/joemccormickjr/

Company: https://www.babylist.com/

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

I put the newest AI coding models from OpenAI and Anthropic head-to-head, testing them on real engineering work I’m actually doing. I compare GPT-5.3 Codex with Opus 4.6 (and Opus 4.6 Fast) by asking them to redesign my marketing website and refactor some genuinely gnarly components. Through side-by-side experiments, I break down where each model shines—creative development versus code review—and share how I’m thinking about combining them to build a more effective AI engineering stack.

What you’ll learn:

1. The strengths and weaknesses of OpenAI’s Codex vs. Anthropic’s Opus for different coding tasks

2. How I shipped 44 PRs containing 98 commits across 1,088 files in just five days using these models

3. Why Codex excels at code review but struggles with creative, greenfield work

4. The surprising way Opus and Codex complement each other in a real-world engineering workflow

5. How to use Git concepts like work trees to maximize productivity with AI coding assistants

6. Why Opus 4.6 Fast might be worth the 6x price increase (but be careful with your token budget)

Brought to you by:

WorkOS—Make your app enterprise-ready today

In this episode, we cover:

(00:00) Introduction to new AI coding models

(02:13) My test methodology for comparing models

(03:30) Codex’s unique features: Git primitives, skills, and automations

(09:05) Testing GPT-5.2 Codex on a website redesign task

(10:40) Challenges with Codex’s literal interpretation of prompts

(15:00) Comparing the before and after with Codex

(16:23) Testing Opus 4.6 on the same website redesign task

(20:56) Comparing the visual results of both models

(21:30) Real-world engineering impact: 44 PRs in five days

(23:03) Refactoring components with Opus 4.6

(24:30) Using Codex for code review and architectural analysis

(26:55) Cost considerations for Opus 4.6 Fast

(28:52) Conclusion

Tools referenced:

• OpenAI’s GPT-5.3 Codex: https://openai.com/index/introducing-gpt-5-3-codex/

• Anthropic’s Claude Opus 4.6: https://www.anthropic.com/news/claude-opus-4-6

• Cursor: https://cursor.sh/

• GitHub: https://github.com/

Other references:

• Tailwind CSS: https://tailwindcss.com/

• Git: https://git-scm.com/

• Bugbot: https://cursor.com/bugbot

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.

Every Monday, host Claire Vo shares a 30- to 45-minute episode with a new guest demoing a practical, impactful way they’ve learned to use AI in their work or life. No pontificating—just specific and actionable advice.

How to build your own AI developer tools with Claude Code | CJ Hess (Tenex)

Brought to you by:

• Orkes—The enterprise platform for reliable applications and agentic workflows

• Rovo—AI that knows your business

CJ Hess, an engineer at Tenex, walks through how he’s built a custom AI development workflow that lets models handle over 90% of his front-end coding. In the episode, CJ demos Flowy, a tool he built to turn Claude’s ASCII plans into interactive flowcharts and UI mockups, and explains why visual planning dramatically reduces cognitive load compared with text. He shares why he prefers Claude Code for intent-heavy work, how custom “skills” make AI tools compound over time, and why pairing Claude for generation with GPT-5.2 Codex for review produces better code than either model alone.

Detailed workflow walkthroughs from this episode:

• How I AI: CJ Hess on Building Custom Dev Tools and Model-vs-Model Code Reviews: https://www.chatprd.ai/how-i-ai/cj-hess-tenex-custom-dev-tools-and-model-vs-model-code-reviews

• Implement Model-vs-Model AI Code Reviews for Quality Control: https://www.chatprd.ai/how-i-ai/workflows/implement-model-vs-model-ai-code-reviews-for-quality-control

• Develop Features with AI Using Custom Visual Planning Tools: https://www.chatprd.ai/how-i-ai/workflows/develop-features-with-ai-using-custom-visual-planning-tools

Biggest takeaways:

1. Claude Code excels at “intent understanding” compared with other models. While CJ acknowledges that GPT-5.2 might be “smarter,” he finds Claude more “steerable” and better at understanding his intentions. This makes Claude particularly valuable for deep dives into complex coding tasks where nuanced understanding matters more than raw intelligence.

2. Skills are the secret to making Claude work with your custom tools. CJ created specific skills that teach Claude how to generate proper JSON for Flowy, with separate skills for flowcharts and UI mockups. These skills evolve alongside his tools, creating a continuously improving ecosystem that makes Claude more powerful for his specific needs.

3. Use model-to-model comparison to improve code quality. CJ uses both Claude (for generation) and Codex (for review) in his workflow. While Claude excels at building features quickly, Codex is better at identifying code smells, inconsistencies, and potential refactoring opportunities. This dual-model approach creates better code than either model could produce alone.

4. Visual planning reduces cognitive overhead compared with text. Even when Claude’s ASCII diagrams contain the same information as Flowy visualizations, CJ finds it much easier to evaluate and approve visual mockups. This highlights how AI tools should adapt to human cognitive preferences rather than forcing humans to adapt to AI output formats.

5. AI can handle more than 90% of front-end coding tasks. CJ says he “hasn’t written a single line of JavaScript or HTML in three months,” instead managing “teams of AI” to write code.

6. “Living dangerously” with AI permissions is increasingly viable. CJ uses an alias named “Kevin” for Claude with bypass permissions, noting that with proper Git safeguards, the risks are manageable.

▶️ Listen now on YouTube | Spotify | Apple Podcasts

If you’re enjoying these episodes, reply and let me know what you’d love to learn more about: AI workflows, hiring, growth, product strategy—anything.

Catch you next week,
Lenny

P.S. Want every new episode delivered the moment it drops? Hit “Follow” on your favorite podcast app.

CJ Hess is a software engineer at Tenex who has built some of the most useful tools and workflows for being a “real AI engineer.” In this episode, CJ demonstrates his custom-built tool, Flowy, that transforms Claude’s ASCII diagrams into interactive visual mockups and flowcharts. He also shares his process for using model-to-model comparison to ensure that his AI-generated code is high-quality, and why he believes we’re just at the beginning of a revolution in how developers interact with AI.

What you’ll learn:

1. How CJ built Flowy, a custom visual planning tool that converts JSON files into interactive mockups and flowcharts

2. Why visual planning tools are more effective than ASCII diagrams for complex UI and animation workflows

3. How to create and use Claude Code skills to extend your development environment

4. Using model-to-model comparison (Claude + Codex) to improve code quality

5. How to build your own ecosystem of tools around Claude Code

6. The value of bypassing permissions in controlled environments to speed up development

Brought to you by:

Orkes—The enterprise platform for reliable applications and agentic workflows

Rovo—AI that knows your business

In this episode, we cover:

(00:00) Introduction to CJ Hess

(02:48) Why CJ prefers Claude Code for development

(04:46) The evolution of developer environments with AI

(06:50) Planning workflows and the limitations of ASCII diagrams

(08:23) Introduction to Flowy, CJ’s custom visualization tool

(11:54) How Flowy compares to mermaid diagrams

(15:25) Demo: Using Flowy

(19:30) Examining Flowy’s skill structure

(23:27) Reviewing the generated flowcharts and diagrams

(28:34) The cognitive benefits of visual planning vs. text-based planning

(31:38) Generating UI mockups with Flowy

(33:30) Building the feature directly from flowcharts and mockups

(35:40) Quick recap

(36:51) Using model-to-model review with Codex (Carl)

(41:52) The benefits of using AI for code review

(45:13) Lightning round and final thoughts

Tools referenced:

• Claude Code: https://claude.ai/code

• Claude Opus 4.5: https://www.anthropic.com/news/claude-opus-4-5

• Cursor: https://cursor.sh/

• Obsidian: https://obsidian.md/

• GPT-5.2 Codex: https://openai.com/index/introducing-gpt-5-2-codex/

• Google’s Project Genie: https://labs.google/projectgenie

Other references:

• Mermaid diagrams: https://mermaid.js.org/

• Figma: https://www.figma.com/

• Excalidraw: https://excalidraw.com/

• TypeScript: https://www.typescriptlang.org/

Where to find CJ Hess:

LinkedIn: https://www.linkedin.com/in/cj-hess-connexwork/

X: https://x.com/seejayhess

Where to find Claire Vo:

ChatPRD: https://www.chatprd.ai/

Website: https://clairevo.com/

LinkedIn: https://www.linkedin.com/in/clairevo/

X: https://x.com/clairevo

Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email jordan@penname.co.