* Maven - Get a $675 discount off Gabor’s course with my code
* Amplitude - The market-leader in product analytics
* Testkube - The leading test orchestration platform
* Land PM Job - My 12-week AI PM + Job Search Course starts Monday!
* Product Faculty - Get $550 off their #1 AI PM Certification with code AAKASH550C7
Today’s episode
Here’s the problem with most Claude Cost demos: they stop at the prototype.
Nobody shows what happens next. You try to add a second feature. The first one breaks. The styling reverts to default. The code is so tangled that you spend more time debugging than you saved by generating.
Gabor Mayer showed me what happens when you stop treating Claude Code like a magic prompt box and start treating it like a team.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit
He is a PM at Google. He has not written production code in 15 years. But over the past several months, he has been building real mobile apps using 21 specialized Claude Code agents. Not prototypes that live in a demo. Apps that are on the App Store.
In today’s episode, he walked through the entire workflow live and share all the resources free.
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
Do you want to become an AI PM? I’ve created a course for you. Starts next week.
Newsletter deep dive
Thank you for having me in your inbox. Here is the complete guide to building a full AI development team in Claude Code:
* Why one-prompt vibe coding fails
* The 21-agent team architecture
* The spec-first workflow
* From design to code without touching either
* What changes when PMs actually build
Save this. The full 10-step playbook on one page. Everything below is the why and how behind each step.
1. Why one-prompt vibe coding fails
Every PM I know has built something with Bolt, Lovable, or Replit. The prototype looks great. It runs. It impresses people in a Slack message.
Then you try to ship it to real users. And you hit a wall.
Blocker 1 - Context compression silently destroys your spec
This is the failure mode that nobody talks about in tutorials. When you give one agent one massive prompt, the model compresses context. Details get dropped. Not randomly. Strategically. The model decides what is “important” and what is not.
In the episode, Gabor defined a complete color palette. Oranges, neutrals, specific accent tones. The agent received everything. The output used none of it. The layout was there. The structure was solid. But every color was a default.
The reason is straightforward. When the context window is full, visual styling details are lower priority than functional logic. So the model drops them. Silently. Without warning. Without an error message. You just get generic output and wonder what went wrong.
The fix is not better prompts. It is context engineering. Smaller, scoped tasks. Each agent gets only the context it needs for its specific job. The designer agent gets the brand guideline. The CTO agent gets the architecture spec. Neither gets the full 50-page document.
Blocker 2 - AI-generated code compiles but is not maintainable
A Reddit comment that hit home for Gabor -
“Vibe coding is just the rebranding of unmaintainable, low-quality source code.”
This is the real prototype-to-production gap. The code works today. You can demo it. You can push it to TestFlight. But the moment you touch it to add a feature, three other features break. No naming conventions. Circular references between modules. Zero comments explaining why anything was built the way it was.
The fix is a dedicated code quality agent. Gabor calls his the Spaghetti Agent. It runs after every sprint and checks naming conventions, circular references, comment coverage, and structural debt. When he ran it on his codebase for the first time, it caught issues he never would have found manually.
If you are building anything beyond a one-off demo, this agent is not optional. I covered similar quality patterns in my AI testing guide and my AI evals deep dive.
Blocker 3 - No dependency mapping means cascading failures
When you build without organizing work into sprints, agents try to build features that depend on code that does not exist yet. Front-end components reference API endpoints that have not been created. Database queries call tables that have not been defined.
The Atlassian MCP currently cannot create sprints directly in JIRA. That is a real limitation. Gabor uses tags as a workaround. He tags tickets as Sprint 1, Sprint 2, Sprint 3 and maps dependencies between them manually before starting the build. Without this step, the entire multi-agent workflow falls apart.
Every PM who has gone from prototype to production with AI agents has hit at least one of these blockers. The ones who shipped figured out the workarounds. The ones who quit assumed the tools were the problem.
Here is what the three blockers look like side by side, and what flips the moment you stop one-prompting and start running a team.
2. The 21-agent team architecture
You do not need 21 agents to start. Three will get you surprisingly far. But understanding the full architecture shows you where the complexity lives and which roles to add as your projects grow.
Here is the full roster: four clusters, 21 roles, and the markdown file pattern that makes them portable across every project you build next.
2a. The core agents every PM needs
The System Analyst is the linchpin. It breaks down product requirements into technical specifications. It asks clarifying questions one at a time. It documents decisions in Confluence. It creates tickets in JIRA. Without this agent, every other agent operates on incomplete context.
In the episode, the system analyst asked 14 clarifying questions before a single line of documentation was written. Vector DB choice. Usage limit mechanics. Conversation history handling. Search fallback strategy. API provider. Minimum iOS version. Screen count. Naming conventions. Each question one at a time so the answers stay deep.
The prompt pattern that makes this work -
“Please act like a good system analyst. Ask clarifying questions until you have a complete and comprehensive understanding. Ask questions one at a time. Do not start writing documentation until all questions are answered.”
Two critical instructions. “One at a time” prevents the agent from dumping 25 questions at once. “Do not start writing” stops it from jumping ahead before the spec is complete. Different LLMs have different tendencies. Some love to start coding instantly. You need to explicitly constrain them. This is the same principle behind the prompt engineering techniques that work across any AI tool.
The Spaghetti Agent handles code maintainability. Naming conventions. Circular references. Comment quality. Structural debt. Born from that Reddit comment. When Gabor ran it on his codebase for the first time, it caught problems he never knew existed.
The UX Flow Architect creates clickable prototypes using Figma’s built-in prototyping arrows. This is a small but important detail. The early versions of this agent placed visual drawn arrows between screens instead of using Figma’s actual prototyping connections. The prototype looked like it had navigation. But when you clicked play, nothing happened. It took months of iteration to fix.
Each agent has a specific Claude Code agent markdown file that defines its role, its constraints, and its interaction patterns. The setup mirrors how you would build a Claude Code Team OS for a human team.
2b. The real blockers nobody warns you about
The Figma MCP color problem. When you connect Claude Code to Figma through the MCP and pass it your full specification, the screens look structurally correct but the colors are wrong. Not slightly wrong. Completely wrong. The model compressed the context and dropped your entire visual identity. The fix is to pass the brand guideline as a separate, focused input to the Designer Agent. Never bundle it with the functional spec.
The Atlassian MCP sprint limitation. The MCP currently cannot create sprints directly in JIRA. Gabor uses tags as a workaround. Sprint 1, Sprint 2, Sprint 3. It works. But it means dependency mapping is a manual step in the system analyst prompt, not an automated feature.
The consumer app vs Claude Code gap. An agent role you set up in the Claude consumer app does not automatically transfer to Claude Code. You need to define agents separately in both environments. The system analyst in your consumer app conversation is a different instance from the system analyst in your Claude Code agent folder. Your AI PM stack needs to account for this separation.
The $200 Max plan economics. On the Max plan, a major build session uses roughly 10% of your monthly allocation. That means you get about 10 full build sessions per month. For a side project, that is plenty. For a production workflow with daily iterations, you need to be deliberate about when you run multi-agent sprints.
2c. Why reusable agents beat fresh setups
Every painful lesson, every edge case fix, every API workaround gets encoded into the agent markdown file. The next project starts from a position of strength. The Spaghetti Agent that took weeks to calibrate on project one is immediately useful on project two. The UX Flow Architect that took months to stop drawing fake arrows works correctly from day one on every subsequent project.
This is the compound interest of building with agents. The first project is slow. The second is faster. By the fifth, your agent team is genuinely effective. Gabor’s Maven course walks through the full setup at maven.com/gabor/productbuilder.
The 21 agents are not the point. The point is that every role on a software team can be replicated by a scoped, reusable AI agent. Start with three. Add roles when you hit friction.
3. The spec-first workflow
Most tutorials start with the terminal. Open Claude Code. Start prompting. Start coding.
That is backwards. The workflow that actually ships production apps starts in the consumer app. On your phone. Possibly while walking your dog. The process maps cleanly to the PM OS framework that works for any complex project.
3a. Define the system analyst role first
Before you describe your app, you ask the LLM to define what a good system analyst does. This creates a behavioral framework that the agent will follow for the rest of the conversation.
The prompt -
“What is the difference between a good system analyst and a bad system analyst in a software development team? Be as detailed as possible.”
The response gives you a blueprint. Requirement elicitation. Stakeholder management. Process modeling. Dependency documentation. You then instruct the agent to act like a good system analyst.
This is the same principle behind AI agents for PMs. Define the role explicitly before assigning the task. It works in Claude Cowork the same way it works in Claude Code.
3b. Dictate, do not type
This is where superwhisper changes the game. In the episode, the app specification was dictated in a single long monologue. Technology stack (Flutter + Firebase). Data storage rules (device-only, no server-side user data). API key security (Firebase Secret Manager, never exposed to front-end). Usage limits (20,000 word cumulative cap with escalating cooldowns). Tone of voice (friendly but firm, like a 20-year referee friend). Vector database configuration (Vertex AI embeddings for IIHF rulebook and Situation Book).
Typing that specification would have taken 30 minutes and produced half the detail. Dictating it took five minutes and captured every nuance. The longest dictation prompt in the history of this podcast.
Here is the actual prompt, the five-step workflow it kicks off, and the two-word constraint - “one at a time” - that stops the agent from face-planting.
The key rule - even if you ramble, even if you are not perfectly concise, the LLM will understand. You lose nothing by over-specifying. You lose everything by under-specifying. This applies whether you are building a prototype or shipping to production.
3c. Documentation before design
The system analyst creates the full Confluence documentation before any design or code begins. Product overview. Technical architecture. AI agent specification. Data flow diagrams. API endpoint mapping.
Without documentation, every agent operates on partial context. With documentation, every agent operates on the same source of truth. I covered this exact approach in my PRDs guide. The principle is identical whether your team is human or AI.
The boring part of building is the specification. The exciting part is watching agents create screens and write code. But if you skip the boring part, the exciting part produces garbage. The PMs who understand product strategy already know this.
4. From design to code without touching either
Once the specification is locked, the workflow shifts from the consumer app to three parallel tracks. This is where the 21-agent architecture pays off and where most of the real-world friction surfaces.
Three tracks - design, tickets, build - running in parallel into four sprints. 72 minutes from idea to App Store submission. Here is the map.
4a. Design through Figma Make and Claude Code
Start in Figma Make. Go to Spotted in Prod. Take screenshots of apps you admire. Feed those into Figma Make to create a brand guideline. Typography. Color palettes. CTA buttons. Error states. Transitions.
In the episode, two inspiration images produced a full brand guideline. One of them was a photo of a laptop cover. Figma Make derived custom colors from the image without manual hex entry.
Claude Code then used the Figma MCP to build actual screens in Figma based on that style guide. Five screens appeared in real time. Each one matching the brand guideline. The Chrome DevTool MCP lets Claude Code visually verify designs in a browser, catching visual bugs the Figma MCP alone cannot detect.
4b. Tickets with the full team review
The system analyst creates JIRA tickets. The entire agent team reviews every ticket before development starts. This is the step that separates production builds from demo builds. Same product launch discipline, different toolchain.
Designer agent verifies screenshots are attached. Test Architect ensures test coverage. Spaghetti Agent sets naming expectations. Product Council confirms data storage policies. CTO Agent validates architecture. This maps to the AI observability principles I wrote about previously.
4c. Sprint execution with the dependency mapping workaround
Tickets organized into sprints using tags (Atlassian MCP workaround). Dependencies mapped. Database setup in Sprint 1. API in Sprint 2. Front-end in Sprint 3. Integration in Sprint 4.
“Claude, start building. Go for Sprint 1. Once done, Sprint 2, then Sprint 3, and so on. If you have any questions, ask.”
Multiple agents work in parallel. The coding phase is the fastest part. On the $200 Max plan, roughly 10% per session.
Everything before the code is the hard part. Once those are right, the code practically writes itself. This is true whether you are shipping to production as a PM or managing an engineering team.
5. What PMs gain by building with agents
If agents can spec, design, code, and test, what is the PM actually doing?
Making product decisions. The tools just got absurdly faster.
Gain 1 - Firsthand understanding of agent behavior
When you interact with agents daily, you develop intuition for context window limits, hallucination patterns, and compression behaviors. That intuition directly improves your roadmap decisions. You stop over-scoping agent features because you know where agents break down. You stop under-investing in evals because you have seen what happens without them.
Gabor has not written production code in 15 years. But he now understands agent behavior better than most PMs who have only read about it. That understanding compounds across every product decision.
Gain 2 - A portfolio that proves competence
A working app on the App Store is undeniable proof. Password-protect a section showing the build process. Confluence docs. JIRA tickets. Agent architecture. That portfolio item says more than any certificate. It says you shipped.
Gain 3 - Iteration speed that compounds
The first build is the hard part. The UX Flow Architect alone took months. The Spaghetti Agent needed weeks of tuning.
But once v1 ships, everything accelerates. New features take a morning. The reusable agent files carry forward every lesson. The PM who has shipped one app can ship the next in a fraction of the time. Not because the tools are better. Because their agents are better.
Stack those three gains over a year and the gap between PMs who build and PMs who watch stops being a gap. It becomes a moat.
You do not need to know how to code. You need a willingness to understand how software works and the patience to specify before you build. If you want to get started, my Claude Code guide walks through the full setup.
How to Become a "Builder PM" with n8n, Claude Code, and OpenClaw | Mahesh Yadav (ex-Google, AWS, Meta, Microsoft; Founder LegalGraph AI)
Monday, April 20, 2026 • Duration 01:36:50
Today’s episode
LinkedIn just changed the title of its product managers to product builders.
What does it even mean to be a “builder PM”?
Well, tools only get you so far. Learning Claude Code is helpful, but means nothing if you don’t have an understanding of the underlying first principles.
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
I’m giving a free talk on how to get interviews at the top AI PM companies on Thursday, April 23rd 2026 @ 9:00AM PDT. Grab your seat.
* NayaOne - Airgapped cloud-agnostic sandbox to validate AI tools faster
* Product Faculty - Get $550 off their #1 AI PM Certification with my link
----
Key Takeaways:
1. Builder PM defined - A builder PM talks to customers, figures out what to build, and ships the first version to 10 customers without talking to any developer. The skill is knowing what to build, not knowing how to code.
2. Four agent components - Every agent that works has intelligence (model), tools (actions), memory (session context), and knowledge (your company data). Every agent that disappoints is missing at least one.
3. n8n for foundations - n8n is the best learning tool because you visually see every component of the agent architecture as separate nodes. Build your first multi-agent system and evaluation pipeline here.
4. Claude Code ate three company types - Context companies, action companies, and evaluation companies all got replaced by one agentic loop inside Claude Code. The three pieces collapsed into one tool.
5. Computer control is the real unlock - File system access plus bash commands equals full laptop capability. This is why Claude Code went from coding tool to work operating system.
6. Long-horizon jobs changed the game - AI agents went from 3-minute tasks to 3-6 hour sustained jobs in six months. This turns Claude Code from assistant to autonomous worker.
7. Continuous learning loops - Build a second agent that watches your corrections to the first agent's work. After five repeated patterns, it proposes a skill update. Your tools get better every day.
8. OpenClaw pattern - Delegation through existing channels, full machine sandboxing, model-agnostic. Not a product but a pattern that Google and AWS will copy inside their ecosystems.
9. AI PM interviews changed - At L5 and L6, product sense questions are being replaced with live building exercises and system design for AI architectures. Pull out Claude Code during the interview or you are already out.
10. Compensation trajectory - From $120K at Microsoft to $1.3M at Google over 13 years, doubling every 18 months through AI-focused switches. Left because big companies kill innovation with six-week approval cycles.
If you want to advertise, email productgrowthppp at gmail.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
How to Design like OpenAI and Figma
Friday, April 10, 2026 • Duration 53:35
Today’s episode
The design process you learned is already dead.
Most teams still follow the same linear pipeline. Low fidelity to high fidelity to handoff. Sketch it. Spec it. Ship it over the wall. That pipeline was built around a constraint that no longer exists. High fidelity used to be expensive. It is not anymore.
I brought in two people who represent both sides of the new design infrastructure.
Ed Bayes is a member of the design staff at OpenAI. He leads design on Codex, which just crossed 2 million weekly users with usage surging 3X since the start of the year. He spends 70-80% of his time coding. He still calls himself a designer.
Gui Seiz is the Director of Product Design for AI at Figma. He leads design on all their AI features, including the Figma MCP server and Figma Make. His designers are now shipping PRs to production.
* Product Faculty: Get $550 off their #1 AI PM Certification with my link
----
If you are trying to understand the new design workflow, this is the one episode to watch.
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
I’m putting on a free webinar on Behavioral and AI PM interviews. Join me.
----
Key Takeaways:
1. Code vs canvas is a false dichotomy - The best designers use both fluidly. Canvas for exploration, collaboration, and pixel-perfect craftsmanship. Code for interactions, responsive testing, and the last mile of polish. The question is what you are trying to learn, not which tool to commit to.
2. High fidelity is no longer expensive - The entire linear design process existed because building something interactive required engineering resources. That constraint is gone. A functional wireframe takes the same time as a paper sketch.
3. The Codex-Figma MCP makes handoff lossless - Import screens from a running React app into Figma with exact pixel values. Border radius, padding, shadows, all one to one. It is not a screenshot. It is a responsive, editable design artifact.
4. The reverse direction works seamlessly - Make changes in Figma, paste a component link into Codex, and it updates your code automatically. No redline spec, no handoff document.
5. Ed spends 70-80% of his time coding and still calls himself a designer - The medium changed but the mandate did not. Designers are still the voice of the user, still upholding craft. The tools expanded, the role stayed.
6. Figma designers are shipping PRs to production - Teams that six months ago were AI curious are now banging down the door. Monetization designers who never wrote code are building technically complex prototypes.
7. "Prototypes, not PRDs" is the emerging norm - PMs at OpenAI bring working prototypes to design reviews. They ship PRs to stress-test ideas before handing off to engineering.
8. You do not need permission to start - Someone from OpenAI's GTM team built an iOS app with zero experience. Download Codex and build something for yourself tonight.
9. Curiosity is the defining skill for this era - Not code proficiency, not design talent. The AI is an infinitely patient tutor. Ask questions. Build understanding alongside output.
10. Total football is the mental model - Every player can play every position. Roles still have natural spikes. But the tool constraints that enforced rigid boundaries are dissolving.
If you want to advertise, email productgrowthppp at gmail.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
How to build a Team OS in Claude Code with Hannah Stulberg, PM @ DoorDash
Tuesday, April 7, 2026 • Duration 01:10:33
Today’s episode
The way PM teams are trending, one PM is going to support 20 people.
Not just engineers. Designers. Analysts. Strategy partners. GTM. Sales. Support.
You cannot answer everyone’s questions about everything. You cannot be in every Slack thread. You cannot be the bottleneck for context that already exists somewhere in a Google Doc no one can find.
But you can give them a high-context, well-organized repo.
Hannah Stulberg is a PM at DoorDash and a former Google PM. She has spent over 1,500 hours in Claude Code.
She wrote the viral Claude Code for Everything series. Her setup is not a personal productivity system. She has structured her entire team’s context into a shared repo that everyone queries.
Her strategy partner - completely non-technical - puts up pull requests every day. Her engineers query metric definitions without asking the analyst. Her designers pull product context without waiting on a PM.
If you are building a team that runs on AI, this is the episode to watch.
* Amplitude: The market-leader in product analytics
* Product Faculty: Get $550 off their #1 AI PM Certification with my link
----
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
I’m putting on a free webinar on Behavioral and AI PM interviews. Join me.
----
1. Build a Team OS, not a personal OS - A shared repo where every function checks in work. Engineers, designers, and analysts self-serve without asking the PM.
2. Root CLAUDE.md is everything - Doc index, team roster with Slack IDs, channel map. Keep under one page or you burn context every session.
3. Nested indexes save 97% of context - Every folder gets a navigation CLAUDE.md. A customer query used only 3% of the context window.
4. Three token tiers - Always-loaded root (~500 tokens), folder indexes on navigation (200-500), content files on demand (1,000-10,000+).
5. Split analytics by product area - Metrics, queries, schemas separated. Progressive loading prevents waste.
6. Gate launches on repo updates - Feature not shipped until metrics, queries, schemas, and playbooks are checked in.
7. Verified playbooks kill hallucinations - Analyst-audited methodology. Claude follows verified steps instead of inventing its own.
8. Plan mode makes 10x docs - Shift+Tab twice. Five phases: load context, ask questions, build plan, push thinking, review agents.
9. Split long docs across parallel agents - Each writes to a temp file. Orchestrating agent compiles. Prevents context overflow.
10. The flywheel compounds daily - Automate one task, free time, improve the repo. After 1,500 hours still iterating every day.
If you want to advertise, email productgrowthppp at gmail.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
How to Turn Claude Code into an Operating System with Carl Vellotti
Monday, March 30, 2026 • Duration 01:06:48
Today’s episode
Claude Code just hit $2.5 billion in annualized revenue in 9 months.
It is the fastest B2B software product ramp in history.
So why are most people still using it like a chatbot?
This is how most people use Claude Code. Type a prompt and get output. The context fills up. It compacts. You lose everything. You start over.
The top users flipped it. They built skills that interview through a framework before building anything. They use sub-agents that preserve context. They have operating systems where every file, every person, every project has a home.
That shift is what today’s episode is about.
I sat down with Carl Vellotti for the third time. His first episode was the beginner course. His second episode was the advanced masterclass. Together they crossed over a million views across platforms.
Today is the operating system layer. If you are already an 80 out of 100 on Claude Code, this episode will bring you to a 95 out of 100.
This episode covers context management, creating sub-agents to manage your context for you, auto-triggering skills with hooks, trustworthy data analysis with Jupyter notebooks, and building an operating system around it all.
If you are living in Claude Code 8 to 10 hours a day and want to stop fighting the tool, this is the one episode to watch.
* Product Faculty: Get $550 off their #1 AI PM Certification with my link
----
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
I’m putting on a free webinar on Behavioral and AI PM interviews. Join me.
----
Key Takeaways:
1. Context management is the real skill - A single web search eats 10% of your context. Run /context to see what is consuming it. System prompt and MCPs take 10-16% before you type one message.
2. Sub-agents save 20x context - Delegate research to a sub-agent. Same task costs 0.5% instead of 10%. Your main session only gets the summary.
3. Replace MCPs with CLIs - MCPs eat context by existing. CLIs have zero overhead. GitHub CLI, Vercel CLI, Google Workspace CLI are all dramatically more efficient.
4. Powerful skills need zero code - Anthropic's front-end design plugin is just a good prompt. No APIs or tooling. Just rules that tell Claude "do not look like AI."
5. Give Claude self-checking tools - The make slides skill uses Puppeteer to screenshot output, measure overflow, and fix issues before you see them.
6. Repeat prompts for better quality - A Google paper showed pasting a prompt twice helps. Tell Claude to double-check against skill instructions after the first pass.
7. Use hooks to auto-invoke skills - A user_prompt_submit hook matches your words against skill keywords instantly. Zero context cost.
8. Jupyter notebooks solve data trust - Every analysis shows exact code, inputs, and outputs. Traceable and reproducible.
9. Build an operating system - Knowledge folder for people context. Projects folder for task isolation. Tools folder for scripts. CLAUDE.md for identity.
10. The people folder compounds - Connect meeting transcription. After every meeting, update each person's dossier. Every prompt gets more specific over time.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
AI PM at Netflix, Amazon and Meta - Here's How to Become an AI PM (Fundamentals + Job Search)
Monday, March 23, 2026 • Duration 01:12:55
Today’s episode
Stop applying to AI PM jobs until you understand the fundamentals.
That is not gatekeeping. That is the MIT finding. 19 out of 20 AI pilots fail. The #1 reason? Picking the wrong problem to apply AI to.
Not the wrong model. Not the wrong data. The wrong problem.
Jyothi Nookula has spent 13.5 years in AI. 12 patents. AIPM at Amazon (SageMaker), Meta (PyTorch), Netflix (Developer Platform), and Etsy.
She has hired AIPMs at three of those companies. Trained 1,500+ PMs to transition into AI roles.
If you are trying to break into AI PM, this is the one episode to watch.
----
Brought to you by
* Product Faculty: Get $550 off their #1 AI PM Certification with my link
* Amplitude: The market-leader in product analytics
* Pendo: The #1 software experience management platform
* NayaOne: Airgapped cloud-agnostic sandbox for AI validation
* Kameleoon: Prompt-based experimentation for product teams
----
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
If you want my PM Operating System in Claude Code, click here.
----
Key Takeaways:
1. Two types of AIPM roles exist - 80% are traditional PM roles with AI features added on, where the core product existed before AI. 20% are AI native roles where the product IS AI and the value proposition is impossible without it. Know which type before you apply.
2. The AI PM stack has three layers - Application PMs own user experience (60% of roles, easiest entry point). Platform PMs build tools for other builders (30%). Infra PMs build foundational systems like vector databases and GPU orchestration (10%).
3. 19 out of 20 AI pilots fail from wrong problem selection - AI makes sense for complex pattern recognition, prediction from historical data, and personalization at scale. If explainability is non-negotiable, rules exist, data is limited, or speed is critical, start with heuristics.
4. Most teams overcomplicate their AI technique choice - If you can put the problem in a spreadsheet with inputs and an output to predict, traditional ML is the answer. Perception problems need deep learning. Natural language reasoning needs Gen AI. These are not competitors, they are tools in your toolkit.
5. AI products are fundamentally probabilistic - The same input can produce different outputs. AIPMs must think in quality distributions and acceptable error rates, not binary success vs failure. Data is a first-class citizen, not a nice-to-have.
6. Agents decide, workflows follow steps - Workflows have predetermined sequences with deterministic outcomes. Agents receive goals and independently decide which tools to use. The live N8N demo showed identical tools producing completely different execution patterns.
7. Context engineering is the real production skill - Claude Sonnet has a 200K token context window but that fills fast with knowledge bases, conversation history, and real-time data. Every token costs money. Managing what to load and when directly impacts both quality and cost.
8. Follow the hierarchy before fine tuning - Prompt optimisation first, then context engineering, then RAG. 80% of use cases get solved with RAG. Fine tuning should only be considered after exhausting all three.
9. Build products not projects - Launch your AI work, get real users, encounter real breakage. That gives you richer interview material than any course certificate. Build an agent, build a RAG system, and build an app that solves a real problem.
10. PM culture at big tech shapes who you become - Amazon PMs spend 40-50% of time writing PRFAQs and six-pagers. Meta PMs live in experimentation and statistical significance. Netflix PMs operate with full autonomy through context over control. Each teaches something different.
PS. Please subscribe on YouTube and follow on Apple & Spotify. It helps!
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
Evals are the new PRD. Here is the playbook with the CEO of the leader in the space (Ankur Goyal, Founder and CEO, Braintrust)
Friday, March 20, 2026 • Duration 51:57
Today’s episode
Most PMs treat evals like a quality gate. Something you run right before shipping, just to check the box.
That is backwards.
The best AI product teams treat evals as the starting point. They write the eval before the prompt. They iterate on the scoring function before the model. They use failing evals as a roadmap.
That shift is what today’s episode is about.
I sat down with Ankur Goyal, Founder and CEO of Braintrust. It is the eval platform used by Replit, Vercel, Airtable, Ramp, Zapier, and Notion. Braintrust just announced its Series B at an $800 million valuation.
Users are running 10x more evals than this time last year. People log more data per day now than they did in the entire first year the product existed.
In this episode, we build an eval entirely from scratch. Live. No pre-written prompts, no pre-written data. We connect to Linear’s MCP server, generate test data, write a scoring function, and iterate until the score goes from 0 to 0.75.
Plus, we cover the complete eval playbook for PMs:
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
If you want my PM Operating System in Claude Code, click here.
* Product Faculty: Get $550 off their #1 AI PM Certification with my link
----
Key Takeaways:
1. Vibe checks are evals - When you look at an AI output and intuit whether it is good or bad, you are using your brain as a scoring function. It is evaluation. It just does not scale past one person and a handful of examples.
2. Every eval has three parts - Data (a set of inputs), Task (generates an output), and Scores (rates the output between 0 and 1). That normalization forces comparability across time.
3. Evals are the new PRD - In 2015, a PRD was an unstructured document nobody followed. In 2026, the modern PRD is an eval the whole team can run to quantify product quality.
4. Start with imperfect data - Auto-generate test questions with a model. Do not spend a month building a golden data set. Jump in and iterate from your first experiment.
5. The distance principle - The farther you are from the end user, the more critical evals become. Anthropic can vibe check Claude Code because engineers are the users. Healthcare AI teams cannot.
6. Use categorical scoring, not freeform numbers - Give the scorer three clear options (full answer, partial, no answer) instead of asking an LLM to produce an arbitrary number.
7. Evals compound, prompts do not - Models and frameworks change every few months. If you encode what your users need as evals, that investment survives every model swap.
8. Have evals that fail - If everything passes, you have blind spots. Keep failing evals as a roadmap and rerun them every time a new model drops.
9. Build the offline-to-online flywheel - Offline evals test your hypothesis. Online evals run the same scorers on production logs. The gap between them is your improvement roadmap.
10. The best teams review production logs every morning - They find novel patterns, add them to the data set, and iterate all day. That morning ritual is what separates teams that ship blind from teams that ship with confidence.
PS. Please subscribe on YouTube and follow on Apple & Spotify. It helps!
If you want to advertise, email productgrowthppp at gmail.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
The Complete Guide to OpenClaw for PMs [EXCLUSIVE]
Tuesday, March 17, 2026 • Duration 01:40:52
This is a free preview of a paid episode. To hear more, visit www.news.aakashg.com
Today’s episode
Every PM I talk to is using AI the same way. Open Claude. Type a question. Get an answer. Close the tab.
The AI does nothing while you sleep. It forgets everything the next morning. It cannot touch your Slack, your email, your file system.
OpenClaw changes that.
245,000 GitHub stars. 2 million weekly visitors. Peter Steinberger built it, Sam Altman bought it for over a billion dollars. I covered what OpenClaw is and why it matters when it first went viral. Today’s episode goes deeper. A complete, step-by-step installation and five PM automations you can copy.
* Product Faculty: Get $550 off their #1 AI PM Certification with my link
----
If you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakash’s bundle.
If you want my PM Operating System in Claude Code, click here.
----
Key Takeaways:
1. OpenClaw is a proactive AI agent, not a reactive chatbot - Unlike ChatGPT or Claude, OpenClaw runs as a continuous daemon on your machine. It executes tasks at 3 a.m. while you sleep, maintains persistent memory across sessions, and acts autonomously based on scheduled cron jobs.
2. Installation takes three terminal commands - NPM install, openclaw onboard, and hatch the bot. If you do not see red text in the terminal, the installation worked. Yellow warnings are normal and safe to ignore.
3. The Slack integration has one critical step everyone misses - Every time you change bot permissions in the Slack API console, you must click Reinstall to Workspace. Without this step, no permission changes persist and the bot appears broken.
4. The workspace docs folder is your team's knowledge base - Drop PRDs, FAQs, and product docs into the local .openclaw/workspace/docs folder. Any team member can query the entire repository by mentioning the bot in any Slack channel, and the bot can write back to the docs.
5. Cron jobs replace manual PM rituals - Set up a morning stand-up summary that scans Slack channels overnight and posts a brief at 9 a.m. with what shipped, active blockers, and customer complaints. You describe it in English and OpenClaw writes the code.
6. Competitive intelligence runs on autopilot - OpenClaw can monitor competitor websites, reviews, and mentions every 30 minutes and post SWOT analyses to a private Slack channel. It tracks changes over time for trend analysis months later.
7. Voice of customer reports aggregate every feedback source - Connect Slack support channels, email, Google reviews, Reddit, and more. OpenClaw scans every 30 minutes and synthesizes a weekly report automatically.
8. Smart bug routing checks customer tier automatically - OpenClaw reads bug reports, looks up the reporter in a customer CSV, escalates enterprise bugs to engineering immediately, and routes free-tier bugs to design as low priority.
9. Security audit is non-negotiable before going live - Tell OpenClaw to analyze its own security vulnerabilities. It will flag unrestricted file access, disabled firewalls, and missing approval gates. Set up a weekly cron job to run the audit automatically.
10. Local deployment is safest for most PMs - A VPS gives 24/7 uptime but removes your physical kill switch. A dedicated Mac Mini is the most recommended option. Local deployment on your laptop is the safest because the bot sleeps when you close your laptop.
If you want to advertise, email productgrowthppp at gmail.
This CPO Uses Claude Code to Run his Entire Work Life | Dave Killeen, Field CPO @ Pendo
Wednesday, March 11, 2026 • Duration 52:48
Today’s episode
Most PMs start every day like this. Open the calendar. Open the CRM. Open Slack. Open the meeting notes. Open LinkedIn. Piece together what matters. Lose 30 minutes before real work even starts.
That is not how the best PMs are working anymore. The best PMs are running one command in the morning and getting everything they need in five minutes. Their calendar, their deals, their market intel, their career gaps, all pulled together automatically.
That shift is what today’s episode is about.
I sat down with Dave Killeen, Field CPO at Pendo.io. He has worked at BBC, Mail Online, and now runs the field product function at one of the largest product management platforms in the world. He has 25 years in product. Over the last few months, he built a full personal operating system called DEX in Claude Code, open sourced it on GitHub, and it is getting serious traction.
In this conversation, Dave walks through his entire system live on screen. You will see how he runs a daily plan, creates PRDs from a backlog, manages parallel workstreams on a Kanban board, and tracks his career goals, all from one terminal window. And you will learn the three building blocks that make it all work.
1. One command replaces your morning routine - Dave's daily plan slash command pulls from calendar, CRM, Granola, LinkedIn, YouTube, and 120 newsletters in five minutes. No tab switching. No manual assembly.
2. MCP servers are the key to connecting everything - Point Claude at any API documentation with your API key and it builds an MCP server for you. MCP provides structured guardrails that make the AI's behavior consistent and deterministic.
3. Skills, MCP, and hooks are three different things - Skills are plain English job descriptions for what the AI should do. MCP servers are structured integrations for connecting external services. Hooks are triggers that fire at specific conversation moments.
4. Session start hooks make the system compound - Every new Claude Code chat gets injected with weekly priorities, quarterly goals, working preferences, and past mistakes. The AI never starts from scratch.
5. Living markdown files are the compounding mechanism - Every project, person, and company gets a markdown file that accumulates context from meetings, messages, and intel over time. The more you use the system, the smarter every file becomes.
6. You can build a mobile app in 37 minutes - Dave built the full app with Claude and spent more time in Xcode publishing it. The constraint is taste, not building speed.
7. The AI should hold you accountable - Dave's Claude MD file includes "harsh truths for Dave" that the AI wrote after auditing his system. This gets injected into every session to prevent the same mistakes.
8. Career planning should compound like product data - A career MCP server collects evidence, runs gap analysis, and calculates promotion readiness. When review time comes, the evidence is already assembled.
9. Be precise about your goal, not the path - The kindest thing you can do for the AI is give it a very clear destination. Do not tell it how to get there. Let it figure out the most elegant approach itself.
10. Voice-first changes everything - Using Whisperflow or Super Whisper instead of typing fundamentally changes how you interact with Claude. You think out loud. The conversation flows. You build faster.
If you want to advertise, email productgrowthppp at gmail.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
Gemini Gem Masterclass From the Creator Lisa Huang
Thursday, March 5, 2026 • Duration 52:11
Today’s episode
Most PMs are using AI the same way they used Google in 2005.
Type something in. Get something out. Move on.
That is not how the best PMs are using it. The best PMs have stopped treating AI as a search engine and started treating it as a team member. One that already knows their product, their writing style, their strategy. One that does not need to be briefed from scratch every single time.
That shift is what today’s episode is about.
I sat down with Lisa Huang, SVP of Product at Xero, an $18 billion finance platform. She built the AI assistant for the first generation Meta RayBan smart glasses. She created Gemini Gems at Google. She has been an AI PM at Apple, Meta, and Google - three of the most demanding AI product environments in the world.
She gave us a masterclass across Gemini Gems, building AI into hardware, running AI agents at scale inside a financial product, and what the AI PM career actually looks like from here.
In today’s episode, we discuss across three topics.
* How to build Gemini Gems and AI projects that actually work.
* What she learned building AI into a wearable device.
* What the future of the AI PM career actually looks like.
1. Stop briefing your LLM from scratch every time - Gemini Gems hold your context permanently. Your role, your company strategy, your writing style. Build it once and it already knows everything the next time you open it.
2. Every PM needs 3 Gems - A writing clone trained on your PRDs and emails. A product strategy advisor loaded with your company docs and competitor analysis. A user research synthesizer that ingests raw transcripts and surfaces key themes.
3. Vague instructions are the number one mistake - "Help me write better" gets you nothing. Write a full page of context. Your role, your audience, your format preferences. The more specific, the more personalized the output.
4. Gemini Gems vs ChatGPT custom GPTs - OpenAI framed GPTs as an app store ecosystem. Google focused on personal productivity instead. First principles beat copying a competitor's framing, and the GPT store never took off.
5. On-device AI is the future for wearables - Cloud is the default today but once a device is on your face all day, people want their data staying local. Privacy beats performance when the device is that personal.
6. Accuracy is the product in high-stakes AI - LLMs out of the box are not great at math, accounting, or tax. Winning agents combine deep domain knowledge with proprietary data that no general-purpose model can access.
7. Measure agents in three layers - Quality first (evals, human annotators, LLM judges). Product metrics second (adoption, retention, CSAT). Business impact third (revenue attribution, ARR). Skip to layer three without the foundation and you are measuring on sand.
8. AI will not replace PMs - it will replace the execution work. Writing PRDs, creating mocks, managing roadmaps. What stays is product judgment. The ability to look at ambiguous signals and say this is the right bet and here is why.
9. The PM role is becoming a hybrid - PM to engineer ratios will compress. The expectation is that PMs also build. Not just spec and hand off, but prototype, design, and code enough to show what they mean. The tools to do this exist right now.
10. Your company's permission is not required - Most companies are not fine-tuning models. They are using the same consumer tools you already have. Build Gems. Build projects. Build small AI products with your personal data. There is nothing stopping you.
If you want to advertise, email productgrowthppp at gmail.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.news.aakashg.com/subscribe
Related Shows Based on Content Similarities
Discover shows related to The Growth Podcast, based on actual content similarities. Explore podcasts with similar topics, themes, and formats, backed by real data.
