“Then voice assistants came, then then there was Alexa and all that stuff”
Alexa
By Amazon
4 recommend/use · 9 sourced episodes
Every sourced reference
Short attributed excerpts only. Timestamps are approximate.
“Ash was an early AI researcher at Alexa and Microsoft and has published over 35 research papers.”
“in my car right now there's Alexa built in, which is great cuz the kids can ask for songs”
“our three-year-old is now capable of just conversing with Alexa Plus”
“if you look at like Alexa and like Siri and Google assistant”
“they have full conversations with chat GPT and Alexa and everything else.”
“the Kindle AWS Alexa Echo um Prime”
“It's produced, you know, it's produced um Kindle, it produced Alexa”
“led the team responsible for growing Alexa internationally”
Related frameworks
Ambitious Retry (Pass@N) Tool Use
Get more from AI tools by asking for the ambitious change and fully restarting on failure instead of hammering the same attempt
Architect-Then-Delegate AI Product Building
Use AI to prototype and fill scoped sub-segments, but architect and lock the system yourself — don't cognitively surrender.
Autonomy at the VP Level
Place decision autonomy at the VP layer — not the leaves, not the top — to maximize thinking without producing chaos.
Benefit = (Volume × Quality) / Time
Measure AI-building impact as experimentation volume times quality, divided by time to launch
Bezos's Three Tests for a New Bet
Big idea, right owner, legitimate plan — an idea needs all three or it dies
Build for the Exponential (Skate to Where the Puck Is Going)
Build products for the model that arrives in 6-12 months, betting that today's 20%-working features hit 100%
Centralized vs. Decentralized Org Choice
Pick your org model from your product strategy: speed (Amazon) or unified experience (Apple), and accept its cost.
Constitutional AI Self-Critique Loop
Align a model to written principles by having it critique and rewrite its own outputs, then train on the fix
Continuous Calibration, Continuous Development (CCCD)
A CI/CD-style loop for non-deterministic AI: scope, evaluate, deploy, then calibrate against surprises.
Decompose-and-Ensemble
Break a broad problem into specific tasks, then solve each with a specialized model in an ensemble.
Don't Wait for the Re-Org
If you're waiting for a formal declaration to build differently, you're already too late
Eval-Driven Product Development
Design the test for your AI product alongside the product, then hill-climb the model against it.
Evals-Plus-Production-Monitoring Dual Feedback Loop
Reject the false dichotomy: evals catch what you know, production monitoring catches what you don't.
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.
Fault-Tolerant User Interfaces
Design your UI to match the real hit-rate of your machine learning, not a fantasy of 100% accuracy.
Golden Samples Over Full Corpus
Curate a small set of gold examples for your AI, don't dump your whole knowledge base on it
Have Backbone, Disagree and Commit
Dissent is an obligation, not a right — and commitment only starts once you've been genuinely heard.
Input Metrics Over Output Metrics
Stop managing the revenue number. Instrument the controllable customer-experience inputs that produce it.
Marginal-Impact Reasoning Under Tail Risk
Prioritize catastrophic low-probability risks by expected value and by how few people are already working on them
Meet-The-Customer-Where-They-Are Marketing
Put your product into the customer's context and re-tune the message for each adopter tier and market.
Micromanage the Decision, Not the Operation
Sweat only the few details that matter for the customer — control the decision, delegate the doing.
Model Maximalism
Build at the edge of what models can barely do — the next model will make it sing.
NLX: Designing the Natural-Language Interface
Conversation is an interface with real constructs — design it, don't just let the model lead
Opinion-Based Decisions for 1.0 Products
For a first-of-its-kind product, empower a tiny 'tastemaker' team to make gut calls — data can't yet exist.
Pain-Plus-New-Technology Idea Selection
Judge what's worth building by pairing a habituated pain with a newly-arrived technology that can finally solve it.
Poor Man's Fine-Tuning (Few-Shot + Role Priming)
Steer a model with in-prompt examples and a role identity instead of a full fine-tune.
Prompt Sets Are the New PRD
Communicate product ideas by building the prototype, not writing the doc — demos before memos
Reason About It Like a Human
To design or debug AI behavior, ask what an equivalent human would do in the same situation.
Recall vs. Discovery Interface Split
Match interface density to user intent: dense grids for recall, low-density sound-rich feeds for discovery.
Resting in Motion
Treat the busy, working state as your normal baseline so heavy long-term work stays sustainable
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 Operating Model
Lightweight bottoms-up planning, PM-light teams, and iterative deployment in a fast-moving domain.
Single-Threaded Leadership
Kill resource contention by giving one leader one mission and every function they need to deliver it.
Solve Before Scale
Run in a distinct 'solve' mode before you ever switch to 'scale' mode
Structural Conditions for Innovation
Companies that say they tolerate failure but punish it in comp and career don't get innovation. Change the structure.
The Agency-Control Autonomy Ladder
Ship AI in graduated versions, trading human control for machine agency only as trust is earned.
The AI Success Triangle
Successful AI adoption is a people problem first: great leaders, good culture, and technical progress.
The Bar Raiser Hiring Process
Put a trained, independent standard-keeper with veto power on every interview loop — and let them run the debrief.
The Career-Stage PM Skill Stack
Master three skills as a junior PM, three different ones as a senior — not all twelve at once
The Cascading Metrics Review
One hour a week of public metric probing that forces every leader to actually know their business
The Economic Turing Test
Measure transformative AI by whether you'd hire an agent for a real job without knowing it's a machine
The Five Traits to Keep, Automate the Rest
Protect where humans shine — vision, empathy, communication, creativity, judgment — and automate everything else
The Full-Stack Builder Model
Empower one builder to take an idea to market end-to-end, regardless of role or team
The Operator's Manual (Answer First)
Answer the question, then explain — or shut up. The tactical checklist for PM communication
The Pantry Test
If the idea starts with the ingredients you already have, you are not working backwards
The Three Dimensions of an Agent
Score any 'agent' on autonomy, complexity, and natural interaction — each a spectrum
The Three Generations Rule
Make the product, fix the product, then fix the business — innovation almost never lands in one generation.
The Two-of-Three Inflection Test
Only build a new zero-to-one product when at least two of three inflections line up
The Working Backwards PR/FAQ Process
Start every new product from the customer's problem, written as a press release, before any constraint enters the room.
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.
Three Worlds Theory of Change
Decide what to actually do about a hard problem by naming the pessimistic, optimistic, and pivotal worlds you might be in
Trust as Currency
Trust is built by repeatedly setting and meeting expectations — and it buys you resources
Truthful Storytelling Through Relentless Refinement
Sell the why, not the what — and refine the story out loud on real listeners until it comes off effortlessly.
Update Your Priors: The AI Capability Arbitrage
Re-test what AI can do today and demand more of it — old scar tissue is leaving alpha on the table
Vertical AI Opportunity Finder
Build where the winning data is locked behind a company or industry's walls — foundation models won't go there.
Working Backwards (Problem-First Product Development)
Start from the customer problem, not the ingredients you happen to have in the pantry
People in these episodes
Related resources
- 37signals
- 80,000 Hours
- A Fire Upon the Deep
- A Gentleman in Moscow
- A Spy Among Friends
- AG1
- AI 2027
- Airbnb
- Amazon
- Amazon Echo
- Amazon Prime
- Amazon Web Services
- Amazon Working Backwards process
- Anthropic
- App Store
- Apple
- Apple Pay
- Apple Podcasts
- Apple Terms of Service
- Area 120
- Artificial Analysis
- Ask Gib
- AssemblyAI
- AWS
- BigQuery
- Black Product Managers Network
- BlackBerry
- Booking.com
- Build: An Unorthodox Guide to Making Things Worth Making
- Cable Cowboy
- caffeinate
- CallRail
- Cameo
- Capital One
- Cerebras Systems
- ChatGPT
- ChatGPT Enterprise
- ChatGPT Voice Mode
- Clair Obscur: Expedition 33
- Claude
- Claude Code
- ClickUp
- Co-Intelligence
- Coda
- Codex
- Cursor
- DALL-E
- Databricks
- DeepMind
- DeepSeek
- Devin
- Discord
- Discover Weekly
- DraftKings
- Drucker on Management
- Dune
- Duolingo
- eBay
- Eco
- Energy and Civilization
- Eppo
- EV News Daily
- Everything Everywhere All at Once
- Figma
- Fin
- Fire Phone
- Flighty
- Gemini
- General Magic
- Getting Real
- GitHub Actions
- GitHub Copilot
- Gödel's Proof
- Good Strategy/Bad Strategy
- Good to Great
- Google Assistant
- Google Docs
- Google I/O
- Google Lens
- GPT-2
- GPT-3
- GPT-4
- GPT-4o
- GPT-5
- Granola
- Greyparrot
- Grok
- Groq
- Halt and Catch Fire
- Helgoland
- Her
- How I Built This
- Humane AI Pin
- Instagram Stories
- Intercom
- iPhone
- iPod
- iTunes
- Khan Academy
- Kindle
- Kurzgesagt – In a Nutshell
- Lenny's Newsletter
- Lenny's Podcast
- Lenny's Podcast website
- Libra
- Linear
- LinkedIn Analyst Agent
- LinkedIn Growth Agent
- LinkedIn Maintenance Agent
- LinkedIn Research Agent
- LinkedIn Trust Agent
- LM Arena
- Loom
- Loonshots
- Lovable
- LucidLink
- Machine Intelligence Research Institute
- Magic Patterns
- Maven
- Messenger
- Meta
- MGM+
- Microsoft
- Microsoft Azure
- Microsoft Clarity
- Microsoft Copilot Enterprise
- Microsoft Teams
- Midjourney
- Mixpanel
- Model Context Protocol
- Montessori education
- Netscape Navigator
- Notion
- OneSchema
- OneSchema FileFeeds
- Open to Work
- OpenAI
- OpenAI API
- Operator
- Our World in Data
- Outlive
- Pantheon
- Perplexity
- Perplexity Pro
- Persona
- Pragmatic Institute
- Prime Video
- Quora
- Raycast
- Recursion
- Remembrance of Earth's Past
- Replacing Guilt
- Researcher agent for work
- Robinhood
- Salesforce
- Saturday Night Live
- Sauce
- Seven Powers
- Seveneves
- Silicon Valley
- Simbe Robotics
- Siri
- Slack
- Something Deeply Hidden
- Song Exploder
- Sora
- Spotify
- Stable Diffusion
- Stanford University
- Star Wars
- Steve Jobs (book)
- Stockholm
- Strava
- Stripe
- Subframe
- Succession
- Superhuman
- Superintelligence
- Sweden
- Ted Lasso
- Tesla
- The Accidental Superpower
- The Alignment Problem
- The Beginning of Infinity
- The Build Collective
- The Case Against Reality
- The Complete Investor
- The Demon in the Machine
- The Effective Executive
- The Fabric of Reality
- The Last of Us
- The Last Question
- The Mystery of the Aleph
- The Wheel of Time
- TikTok
- Top Gun: Maverick
- Twitch
- Uber
- United Nations Universal Declaration of Human Rights
- Vanta
- Visual Studio Code
- Waymo
- What Separates the Top 1% of Product Managers from the Top 10%?
- When Breath Becomes Air
- WhisperFlow
- Why Nations Fail
- Windsurf
- Wispr Flow
- Wix Studio
- Wool Trilogy
- Working Backwards
- Yahoo
- Yellowstone
- YouTube
- Zapier
- Zipp 303 Firecrest
Spot an error or want this page removed? Request a correction or removal.