Product Growth Podcast

• How to Build Multi-Agent AI Systems That Actually Work in Production | Tyler Fisk

  Tyler Fisk built a $1.6 million AI education business in one year.Zero PhDs. Zero Silicon Valley pedigree. Just a systematic approach to building AI agents that actually work in production.While everyone’s vibe coding in ChatGPT, Tyler’s teaching thousands of students to build multi-agent systems for real businesses. Hundreds of production deployments. Actual revenue.Today he’s doing a live build: Taking Apple customer service from idea to working multi-agent system in under 90 minutes.No theory. Pure execution.----Brought to you by:* Maven: Get $300 off Mahesh’s course with my code AAKASHxMAVEN* Vanta: Get $1,000 off AI security & compliance at vanta.com/acos* Testkube:* Kameleoon: Leading AI experimentation platform* The AI Evals Course for PMs: Get $1155 off with code ‘ag-evals’----Key takeaways:1. Stop Vibe Coding: Most teams write one prompt, test twice, ship to production, and hope for the best. Tyler's rule: "We would never put it into production without a human-in-the-loop checkpoint. That's irresponsible." Start with 100% human review, gradually move to 60-70% autonomy.2. Use Meta-Prompting to Build Agents: Tyler built Gigawatt—an agent with 72,000 characters of system instructions that builds other agents. It researches the domain, writes V1 instructions, evaluates itself (scores out of 100), identifies gaps, and rewrites to V2. Goes from 77% to 86%+ quality.3. Build Multi-Agent Architectures: Don't build one agent that does everything. Separate concerns like you'd separate teams. For Apple: Core (expert agent, temp=0, finds facts) + Echo (email agent, temp=0.7, writes responses). Each optimized for its specific role.4. System Instructions Need 7K-9K Tokens: Structure includes Role (job description), Context (business details), Instructions (step-by-step process), Criteria (guardrails), Examples (meta reasoning). Most people write 200 tokens. Tyler writes 7,000-9,000. That's the foundation.5. Temperature Is Your Secret Weapon: Tyler's Toy Story analogy: Imagine an icy peak in a claw machine. Temp=0 (frozen): claw picks from top only—deterministic, precise. Temp=1 (melted): claw grabs anywhere—creative, varied. Match temperature to agent's job.6. Information Hierarchy Prevents Hallucinations: Priority order: RAG database first (scraped company docs), System instructions second (built-in expertise), Web search third (with chain-of-verification). When agents search without verification, they hallucinate.7. Build Complete Workflows: Tyler's 9-step production workflow with 5+ agents: Email arrives → Sentiment analysis (Cinnamon) → Expert research (Core) → Email writing (Echo) → QA loop → Human checkpoint (Slack) → Generative filter → Send → Log to memory.8. Observational Evals Come First: Test 20+ different scenarios manually. Include edge cases and adversarial inputs. Document every failure. Save golden examples. Only after building confidence do you add systematic evals in production.9. Calculate ROI as Labor Cost Reduction: Traditional cost: $460/day (expert time + customer service rep + manager review) = $138K/year. AI cost: $153/day (platform fees + API credits + human review) = $45.9K/year. Savings: $92K annual (67% reduction).10. Emotion Prompting Actually Works: Tyler ends every prompt with "Go get 'em slugger." Based on research: positive reinforcement improves LLM outputs by ~15%. The same psychology that works on humans works on LLMs. "Be nice to your AI. They're gonna have robot bodies soon."----Related ContentPodcasts:* Warp CEO on Profitable AI Agents* Elizabeth Laraki on AI Product Design* Claude Code TutorialNewsletters:* AI Agents: The Ultimate Guide for PMs* How to Build AI Products Right* Ultimate Guide to AI Prototyping Tools----P.S. More than 85% of you aren’t subscribed yet. If you can subscribe on YouTube, follow on Apple & Spotify, my commitment to you is that we’ll continue making this content better.----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

  --------  

  1:41:03

  --------

  1:41:03
• Crash Course in AI Product Design from Google Search + Maps Designer, Elizabeth Laraki

  Today’s EpisodeEveryone’s building AI products wrong.They’re sprinkling AI on top like fairy dust. Adding chat interfaces to everything. Ignoring 70 years of design principles.Elizabeth Laraki was one of 4 designers on Google Search in 2006. One of 2 designers on Google Maps in 2007. She helped create products used by billions—products whose designs barely changed for 15+ years because they nailed it from the start.Today she breaks down exactly how to design AI features that users actually love.----Check out the conversation on Apple, Spotify and YouTube.Brought to you by:* Vanta: Automate compliance, manage risk, and prove trust* Kameleoon: Leading AI experimentation platform* The AI PM Certificate: Get $550 off with ‘AAKASH550C7’* The AI Evals Course for PMs: Get $1155 off with code ‘ag-evals’----Timestamps:00:00:00 - Intro00:01:52 - Elizabeth's background at Google00:04:19 - Google's AI search integration00:06:19 - Designing image & video for AI00:09:44 - AI image expander disaster00:16:05 - Ads00:17:50 - AI safeguards & human-in-the-loop00:18:28 - 3-step AI design process00:31:29 - Ads00:33:25 - Designing AI voice interfaces00:38:25 - Designing beyond chat00:41:52 - AI design tools for designers00:44:49 - Live design: LinkedIn for AI00:57:04 - Google Maps redesign story01:04:14 - Google Maps India landmarks01:10:09 - Where to find Elizabeth01:12:00 - Outro----Key Takeaways1. The Core Design Process Hasn't Changed: Define the product (who, what tasks, what needs), Design it (features, architecture, flows), Build it (UIs, brand). Don't skip to "let's add a chatbot" because you have API access. The fundamentals still apply for AI.2. AI Adds Non-Deterministic Risk: Traditional software is deterministic - click A, get B every time. AI is non-deterministic with unpredictable outputs. Elizabeth's image expander added a bra strap that wasn't in the original photo. Completely unintentional, completely unacceptable.3. Work With Research on Safeguards: Audit training data for bias. Build evals that flag sensitive content (human bodies, faces, private information). Show A/B options for ambiguous cases. Make AI's work visible in the UI so users can scrutinize changes.4. Start With Jobs To Be Done: Don't ask "We have GPT-4, what should we build?" Ask "What painful workflow takes users hours?" Descript mapped video editing lifecycle and baked AI into each job: remove filler words, edit from transcript, create clips, write titles.5. Map User Context, Not Just Needs: ChatGPT voice in car with three kids? Perfect - nobody's looking at screen. Meta Ray-Bans reading Spanish menu item by item? Terrible - should ask "What are you in the mood for?" Same AI, different context requires different design.6. Emerge From Ambiguity First: For "LinkedIn for AI," Elizabeth mapped 4 possible directions, picked Matchmaking, identified AI's unlock (personality patterns vs keyword matching), mapped separate UIs for job seekers and employers. Only then touch pixels.7. Chat Fails for Complex Tasks: Elizabeth tried creating Madrid itinerary in ChatGPT. Every change regenerated everything with new hallucinations. Chat works for Q&A but fails for document creation, visual tasks, multi-step workflows that need persistent editable outputs.8. Make Chat Supporting, Not Primary: Photoshop embeds AI in existing canvas tools. Google Search shows AI summaries inline in normal results. Cove gives canvas with multiple AI conversations in parallel. Chat is a tool, not THE interface.9. Stop Adding AI Sprinkles: Elizabeth: "I can't help but think of this massive container of AI sprinkles everybody's shoving on top." Twitter/X + Grok, Amazon + Rufus, Apple Photos all feel forced. Ask three questions: Is this solving a real problem? Does chat make sense? Can you show your work?10. Google Maps India Innovation: Researched how Indians actually navigate (by landmarks, not street names). Identified which landmarks work (visible from street level like temples, petrol stations). Redesigned entire directions system around that insight. That's design, whether AI or not.----Where to Find Elizabeth Laraki* Linkedin* X (Twitter)----Related ContentPodcasts:What it means to be Design-LedComplete Tutorial to AI Prototyping5 AI Agents Every PM Should BuildNewsletters:Ultimate Guide to Product DesignUltimate Guide to AI PrototypingHow to Work With Design for Success----P.S. More than 85% of you aren’t subscribed yet. If you can subscribe on YouTube, follow on Apple & Spotify, my commitment to you is that we’ll continue making this content better.----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

  --------  

  1:12:12

  --------

  1:12:12
• The Claude Code Tutorial for AI PMs: Why You Need to Use It + How

  Today’s EpisodeClaude Code hit $500 million ARR in four months.Two product managers. Zero marketing dollars. Just pure viral growth.While some PMs are still copying and pasting into ChatGPT, others are orchestrating multiple AI agents that work in parallel, automatically reading files, researching competitors, and building prototypes.Carl Vellotti runs the world’s largest PM Instagram account (55K followers) and has mastered Claude Code better than almost anyone. He’s built his own meme generation system, automated his content workflow, and uses Claude Code for everything from research to prototyping.Today’s tutorial takes you from beginner to Claude Code hero.----Check out the conversation on Apple, Spotify and YouTube.Brought to you by:* Linear: Plan and build products like the best----Key Takeaways1/ Stop Working in Chat WindowsTraditional chat requires manually dragging files one at a time. Claude Code lives in your terminal and automatically reads entire folder structures. The interface was the bottleneck all along.2/ Build Your Knowledge Base FirstCreate four folders: business-info.md for product context, writing-styles/ for different voices, examples/ for past PRDs, meeting-transcripts/ for automatic uploads. One prompt pulls from everything.3/ Use the CLAUDE File for MemoryAdd rules once, they persist forever. "Never commit without asking." "Always use technical writing." Unlike prompts that get lost in context windows, this stays active every session.4/ Save Your Best Prompts as CommandsCreate /meeting-notes, /competitive-research, /prd-review. Save once, reuse forever. No more hunting through old Twitter bookmarks for that perfect prompt.5/ Let Claude Plan Before ExecutingPress Shift+Tab for Plan Mode. Claude creates full execution plan without touching files. You review, catch mistakes, then approve. This one habit prevents 80% of AI disasters.6/ Parallelize Everything You CanNeed to analyze 3 customer interviews? Claude spins up 3 UXR agents working simultaneously. Week of manual work becomes 1 hour with parallel agents.7/ Build Custom Agent PersonalitiesDesigner agent focuses on UX. Engineer agent checks technical constraints. Executive agent evaluates business impact. All three review your PRD simultaneously with specialized perspectives.8/ Use the $37/Month ComboClaude Pro ($17) handles research and writing perfectly. Add Cursor ($20) for heavy coding. You get best models for $37 instead of $200/month Claude Max.9/ Only See Token Usage HereClaude Code shows real-time token consumption and cost. Finally understand what API pricing actually means. No other interface gives you this visibility.10/ Start Simple Then ScaleBegin with one research task using file analysis. Add a custom command. Try parallel agents once. You'll never go back to chat interfaces.----Where to Find Carl Vellotti* Linkedin* X (Twitter)* Instagram----Related ContentPodcasts:Cursor TutorialWindsurf TutorialAI Prototyping TutorialNewsletters:AI Agents: The Ultimate Guide for PMsUltimate Guide to AI Prototyping ToolsHow to Land a $300K+ AI Product Manager Job----P.S. More than 85% of you aren’t subscribed yet. If you can subscribe on YouTube, follow on Apple & Spotify, my commitment to you is that we’ll continue making this content better.----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

  --------  

  1:37:04

  --------

  1:37:04
• The AI PM’s Guide to Building AI Agents, with Warp CEO Zach Lloyd

  Today’s EpisodeAs an AI PM, you’re probably tired of building AI Agents and don’t know how to monetize them.But what if I told you there’s a company adding $1 million ARR every 10 days with their AI agent?Zach Lloyd, CEO of Warp and former Google engineering leader, cracked the code. His terminal-based AI agent has 700,000+ active developers paying real money.This episode is his complete playbook for AI PMs who want to build agents that actually make money.I hope you enjoy this one!----Brought to you by:* Vanta: Automate compliance, manage risk, and prove trust* Kameleoon: Leading AI experimentation platform* Amplitude: The market-leader in product analytics* The AI Evals Course for PMs: Get $1155 off with code ‘ag-evals’----Timestamps00:00:00 - Intro00:01:55 - Interview Begins00:02:02 - Warp's Scale & Growth00:03:08 - The Turning Point00:04:32 - Learn or Get Left Behind00:05:50 - Framework for AI Value00:08:30 - Warp's Development Process00:12:28 - UX Challenges in Agentic Products00:14:53 - Ads00:19:29 - Who's Making Money with Agents00:28:31 - Future Predictions00:29:24 - Ads00:30:26 - Contrarian Takes on AI's Future00:35:44 - 90-Day Roadmap for PMs00:38:33 - Outro----Key Takeaways----Where to Find Zach Lloyd* Linkedin* X (Twitter)* Warp----Related ContentPodcasts:* He built the top AI agent startup* AI Agents for PMs in 69 Minutes* How to Build AI Agents (and Get Paid $750K+)Newsletters:* AI Agents: The Ultimate Guide for PMs* Ultimate Guide to AI Prototyping Tools* How to Land a $300K+ AI Product Manager Job----P.S. More than 85% of you aren’t subscribed yet. If you can subscribe on YouTube, follow on Apple & Spotify, my commitment to you is that we’ll continue making this content better.----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

  --------  

  39:25

  --------

  39:25
• The AI PM's Guide to Security - with Okta's VP of PM & AI, Jack Hirsch

  Today's EpisodeHere's what's happening right now:Someone can clone your voice from a few YouTube videos and call your help desk pretending to be you.AI can build a perfect fake of your login page in minutes.This isn't some distant future threat. Jack Hirsch, VP of Product at Okta, sees this happening every day. Okta protects millions of logins and Jack has a front-row seat to how AI is completely changing cyber attacks.And the scary part is most PMs have no idea this is happening to their products.That's why I brought Jack on the show. He breaks down what's really happening and what you need to know as someone building products in the AI era.----Brought to you by:* Amplitude: The market-leader in product analytics* The AI Evals Course for PMs: Get $1155 off with code ‘ag-evals’* The AI PM Certificate: The #1 AI PM certificate* Kameleoon: Leading AI experimentation platform----Key Takeaways1. Identity is Everything: Over 80% of breaches stem from identity attacks, not device or network vulnerabilities. You cannot get security right without getting identity right - this is the new reality.2. DPRK Infiltration Operations: North Korean agents are passing full interview processes, getting hired, having laptops shipped to device farms, and operating as inside threats within major organizations.3. AI Agents = Security Blindspot: Companies deploy AI agents en masse without treating them as identities requiring access management. JP Morgan's CISO called this out as the biggest current threat vector.4. Help Desk Social Engineering: Attackers use AI voice cloning and deepfakes to impersonate employees calling help desk for password resets, MFA bypasses, and account access - often successfully.5. Session Security Over Time: Authentication degrades after login. Okta focuses on continuous session monitoring and risk signal sharing between security vendors rather than constant MFA prompts.6. T-Shaped Identity Strategy: Deep identity security (phishing-resistant auth, lifecycle management, risk sharing) plus broad integration across all enterprise systems - not just SSO and MFA.7. Cross-App Access Standard: New OAuth standard allows AI agents to inherit user permissions across enterprise apps without individual OAuth dances for thousands of employees.8. Essential vs Discretionary AI: Essential AI (bot detection, fraud prevention) stays always-on. Discretionary AI (log summaries, access reviews) gives customers opt-out control for compliance.9. AI Product Principles: Accelerate don't abdicate, solve real problems before prototyping, ignore AI hype cycle. Use AI as thought partner, not replacement for product judgment and domain expertise.10. Personal Security Stack: Lock credit reports immediately, use password manager with unique passwords, enable passkeys everywhere, lock phone number with carrier PIN to prevent SIM swapping attacks.----Related ContentPodcasts:How to Get a Product Leadership JobHow He Became a Series C VP of Product in 10 Years“Product Management isn’t going to exist in 5 years” - 2x CPONewsletters:The Product Leadership Job SearchThe Product Leader’s Ultimate Guide to Process ChangesProduct Leadership Interviews (GPM, Director, VP): How to Succeed----P.S. More than 85% of you aren't subscribed yet. If you can subscribe on YouTube, follow on Apple & Spotify, my commitment to you is that we'll continue making this content better.----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

  --------  

  1:30:22

  --------

  1:30:22

More Business podcasts

Trending Business podcasts

About Product Growth Podcast