“right around the time that GB4 finished training”
GPT-4
By OpenAI
5 recommend/use · 9 sourced episodes
Every sourced reference
Short attributed excerpts only. Timestamps are approximate.
“we had done the calculation for what it took to to train GP4”
“qb4 came out the the consensus from the world is everything is different”
“when I started with Microsoft started working on on gp4 with Microsoft in like September of last year”
“when gp4 was first launched um I saw a friend of a friend”
“The obvious one is is uh Chat GPT GPT-4 and just playing around with that”
“We realized we can have GPT-4 read all of our docs, store them as embeddings, and then answer questions for developers”
“there's this meme that gpt4 is gonna or five or six or ten is gonna replace product managers”
“where I'm more inclined for PMS to try to get leverage out of something like gpt4 is a lot of the tedious work”
“do you think chat GPT or just say GPT 4 or J GPT 5 GPT 6”
Related frameworks
Alpha Co-Creation Loop
Pick the users who are already pushing the boundary, build with them in a shared channel, ship only when they're thrilled.
Assume the Disruptor Is Right
During an existential threat, assume the disruptor is playing an optimal game and respond with overreaction
Autonomy at the VP Level
Place decision autonomy at the VP layer — not the leaves, not the top — to maximize thinking without producing chaos.
Barrels-and-Ammunition Team Design
Staff each team from the gap, not from a fixed PM/EM/designer template
Centralized vs. Decentralized Org Choice
Pick your org model from your product strategy: speed (Amazon) or unified experience (Apple), and accept its cost.
Chop-It-Up AI Collaboration Loop
Don't hand AI one giant spec — specify a little, review a little, repeat in tight loops.
Complement-the-Frontier Model Strategy
Don't rebuild foundation models — train small specialty models that attack their weaknesses in speed, cost, and niche tasks.
Consumer Subscription Viability Test
Freemium-plus-paid-acquisition consumer subscriptions fail without ~60-70%+ retention or a network effect
Context Is All You Need Prompting
Treat the model as a brilliant stranger with zero context, and supply what a colleague would already know.
Dogfood-Driven Realism
Be the end user, use your product intensely daily, and never ship anything that isn't useful to you.
Draft-First LLM Augmentation
Never ask the model to do your job — write your version first, then have it improve it.
Evals as Articulating Success
An eval is just a clear spec of ideal behavior — the shared language of AI product work
Explain-It-to-Prove-You-Understand-It
Force yourself and others to explain decisions; if you can't articulate it, you probably don't understand it.
Fake the AI Before You Build It
Never train a model for an MVP — prototype the AI's output and test demand first.
Fall-On-Your-Face Taste Calibration
Deliberately push AI past its limits in a safe environment to build a gut feel for what it can do.
Fault-Tolerant User Interfaces
Design your UI to match the real hit-rate of your machine learning, not a fantasy of 100% accuracy.
Founder Intuition vs Team Expertise Handoff
Delegate to hires only once they consistently make better calls than you would — and signal in both directions
Friction Logging
Adopt a specific user's identity, walk your own product end-to-end, and log every point of friction.
Group UX Review and Walking the Store
Taste the soup together — experience the product as a group, log issues live, then debate each one.
Guilty-Pleasure Career Bet
Find the thing you feel guilty getting paid for, do the hell out of it with intensity, and take non-linear risks.
High Agency, High Urgency Hiring Filter
Hire for two traits only — people who see a problem and go, and people who go now.
High-Ceiling Market Test
Attack markets with a high ceiling and easy switching; incumbents win only where there's little left to build.
Hire for Autonomy: References, Real Work, and Default Trust
You'll be in none of the decisions that matter — so hire people you can hand autonomy to, and verify via references.
Internal Tools as a Product (The Crying Octopus)
Run your dev-productivity team like a product team: monthly surveys, hard metrics, and one-click paper cuts.
Knowledge-Work End-State Idea Generation
Pick an area of knowledge work and reason forward to its end state as AI matures — then build for that.
Leap-of-Faith MVP Scoping
Scope the smallest build that tests your riskiest assumption, not a stripped-down product
Measure in Hundreds
If your unit of measurement is one hundred attempts, five failures means you have effectively tried zero times.
North Star Exploration
Turn aimless tech-tinkering into learning by chasing an arbitrary-but-real goal and being stubborn about reaching it.
Recall vs. Discovery Interface Split
Match interface density to user intent: dense grids for recall, low-density sound-rich feeds for discovery.
Robust-Across-Futures Action
When the future is unknowable, choose actions that are ethical across a wide range of scenarios
Run Toward the Hard Use Cases
Don't disable high-stakes uses to avoid downside — engineer them to be great
SaaS-Marketplace Transition Test
A SaaS business can add a marketplace only if it touches its customer's customer and those customers need multiple suppliers
Scenario-Based PM Interviewing
Interview PMs by making them do the real job under real constraints, not recite prepared stories
Separating Change-Aversion from Real Problems in a Redesign
After a redesign, split 'upset because it changed' from 'upset because it's worse' before you react.
Ship Fast Without Breaking Money
Reject the reliability/velocity trade-off: automated gauntlet, ramped exposure, and remediations ahead of roadmap.
Ship-to-Learn Research Allocation
Reserve scarce user research for high-uncertainty, high-leverage problems; ship to learn everywhere else
Ship-to-Learn: The Emergent-Product Loop
When product properties are emergent, launching is how you discover them
The Adjacent-Precedent De-Risk Pitch
Win leadership buy-in for a big AI bet by anchoring it to a past bet that already worked.
The AI PM Upskilling Path
Learn the fundamentals, shadow a research scientist an hour a week, and build one model end to end.
The Best-Practice Inversion Interview
Ask a candidate for a best practice, then for when it would not apply
The Better Tool, Same Problems Lens
Every model leap gets normalised within months — build for the boring future, not the euphoric one.
The Chief Engineer Trade-off Owner
Give one person moral authority over a whole product to keep design coherence in one head
The Constrained-Resource Headcount Test
When the bottleneck isn't people, each new hire is a net productivity loss unless they uplevel everyone.
The Engineer-cation
Leaders clear 3-4 days, join a team as an IC, ship one small feature to production, and log the friction.
The Fixed-Window Pivot Decision
Time-box a full-focus test on the one thing that matters, then let the team name the next bet
The Four Challenges of AI Product Management
Uncertainty, pivots, data scarcity and a broken promo path — the four taxes of the AI PM role.
The Inverted W Planning Process
Teams propose, leaders synthesise, teams adjust, leaders resynthesise, orgs distribute with context.
The Love-Hate Disruption Test
Gauge whether an idea is truly disruptive by how polarized the reactions are, not how positive.
The Maximally Accelerated Question
A forcing question that separates critical path from what can wait
The Model Launch Bar
With probabilistic products, the PM — not the scientist — decides what accuracy is good enough to ship.
The Platform Encroachment Test
Build where the platform's mission says it will never go — the general layer is not yours to own.
The Proxy Goal Ladder
Every metric you chase is a proxy — climb the ladder to the mission before you optimise it.
The Shiny Object Trap (Problem-First AI)
A regular PM ships the right product; an AI PM solves the right problem.
The Sticky Engine of Growth
Model retention by the three, and only three, reasons a customer returns to your product
The Three Network Effects (and Why Social Products Must Evolve)
Classify defensibility as direct, cross-side, or data network effects — and know social products must add the latter two
Think It, Build It, Ship It, Tweak It
Four product phases where spend rises stage by stage — so you must retire risk before the money starts.
Three-Model Triangulation
Apply three different mental models to a problem; agreement across them sharply raises your odds of being right.
Treat Your Course Like a Product
Hypothesise the audience, interview them, iterate the ICP, and run three weeks — not one.
Trust-by-Structure Governance
Embed your promises into the company's structure so they hold even when you're gone
Two-Day Work-Test Hire
Bring finalists on-site for two days to do a real end-to-end project — it screens skill and fit at once.
Two-Mode Prioritization for AI Products
Prioritize backward from model magic AND forward from customer needs
User-Value Thermodynamics
Users are lazy; adoption only happens when total value delivered exceeds total effort spent, ideally by 10x.
What-If Before Why-Not
Evaluate a disruptive idea by imagining its upside first, then treat the objections as your build list.
People in these episodes
Related resources
- A Fire Upon the Deep
- Ahrefs
- Airbnb
- Airtable
- Alexa
- Amazon
- Amazon Prime
- Amazon S3
- Amazon Web Services
- Amplitude
- Andor
- Andrej Karpathy's YouTube channel
- Apple
- Apple Pay
- Apple Podcasts
- Arcade
- arXiv
- Atlassian
- AutoML
- AWS
- Bard
- Blue Apron
- Booking.com
- Box
- Boz's podcast
- Build
- Calm
- Cameo
- Canva
- CareerFoundry
- Casey Accidental
- ChatGPT
- ChatGPT Plus
- Chime
- Chromium
- Claude Sonnet
- ClickUp
- Coding Dojo
- Conductor
- Coursera
- Cursor
- DALL-E
- Discord
- Discover Weekly
- Disney
- DoorDash
- DraftKings
- Dropbox
- Duolingo
- Eco
- Eppo
- Eventbrite
- Faire
- Figma
- Gemini
- General Assembly
- GitHub Copilot
- GitHub Enterprise
- Glitch
- Gmail
- Gödel's Proof
- GoFundMe
- Google Cloud
- Google Docs
- Google Forms
- Google Glass
- Google I/O
- Google Maps
- Google Sheets
- Google Ventures
- Google Workspace
- GPT Store
- GPT-3
- GPT-4o
- GPT-5
- GPTs
- Gran Turismo
- Granola
- Grubhub
- Hacker News
- Halt and Catch Fire
- Harvey
- Headspace
- Helgoland
- Her
- Hex
- High Output Management
- IMVU
- Inspired
- Instacart
- Introduction to Artificial Intelligence
- Invisible Cities
- iPhone
- Jira Product Discovery
- Kicking and Screaming
- Kindle
- Lenny's Newsletter
- Lenny's Podcast
- Lennybot
- Lensa
- Levoit Classic 300S Ultrasonic Smart Humidifier
- Linear
- Link
- Llama
- Long-Term Stock Exchange
- Loom
- Manta Sleep Mask
- Marginal Revolution
- Maven
- Meta
- Microsoft
- Microsoft Clarity
- Microsoft Windows
- Midjourney
- MIT
- NASA
- Netflix
- Notion
- Notion AI
- o3
- OneSchema
- OneSchema FileFeeds 2.0
- OpenAI
- OpenAI API
- OpenAI Prompt Engineering Guide
- OpenTable
- Orkes
- Palo Alto
- Pando
- Patreon
- PayPal
- Perplexity
- Philip Morris
- Poor Charlie's Almanack
- PostHog
- Postmates
- Profit from the Core
- Python
- Ramp
- Reforge
- Retool
- Ruby
- Runway
- Salesforce
- Sanity
- Scaling People
- Scavengers Reign
- Seamless
- Segment
- Seven Powers
- Severance
- Shopify
- Siri
- Slack
- Slow Horses
- Smithsonian Institution
- Snowflake
- Something Deeply Hidden
- SpaceX
- Spotify
- Stable Diffusion
- Stack Overflow
- Stanford University
- Star Wars
- Station Eleven
- Stockholm
- Stranger Things
- Stripe
- Stripe Atlas
- Stripe Billing
- Stripe Connect
- Stripe Sessions
- Substack
- Succession
- Superhuman
- Sweden
- TechCrunch
- Tesla
- The Beginning of Infinity
- The Case Against Reality
- The Complete Investor
- The Dandelion Dynasty
- The Demon in the Machine
- The Design of Everyday Things
- The Download
- The Enlightened Capitalists
- The Fabric of Reality
- The Goal
- The Last of Us
- The Lean Startup
- The Murderbot Diaries
- The Mystery of the Aleph
- The One World Schoolhouse
- The Wasp Factory
- The White Lotus
- Thinking, Fast and Slow
- TikTok
- TLDR
- Toyota
- Twitch
- Uber Eats
- Universal Primer
- Van Westendorp Price Sensitivity Meter
- Vanta
- Venmo
- Visual Electric
- Visual Studio Code
- WAOAW Sleep Mask
- Watershed
- Waymo
- Westworld
- Whatnot
- Where Good Ideas Come From
- Whimsical
- Why We Sleep
- World of Warcraft
- Y Combinator
- Yahoo
- You Look Like a Thing and I Love You
- YouTube
- Zapier
- Zillow
- Zoom
Spot an error or want this page removed? Request a correction or removal.