Lenny's Evil Twin's Podcast
Leonard, Lenny's evil twin, hosts a podcast where he evaluates startups using data from Lenny's five years of experience. He decides whether a startup should be shipped or not, based on his unique perspective.
Episodes
Firma.dev — Leonard wouldn't ship this
Three cents. You send a wedding invitation with more financial conviction than this product charges for a legally binding contract. | Firma.dev's "100x refund" math at three cents an envelope is adorable right now — run it back when they're processing real volume. | The enterprise deals that actually fund a company require exactly the sales cycle they built their whole identity around hating.
Sou
Bond — Leonard wouldn't ship this
Bond's "executive" ICP could be 40 million people. That's not a target audience, that's a census. | Bond's marquee social proof is one testimonial from the CEO of Infinity Constellation — a company I cannot find. Elena Verna puts it plainly: ten customers isn't data, that's a Google Sheet. You don't build a retention thesis on a G-sheet. | Bond is promising more autonomy than the company that inve
Publora — Leonard would ship this
Today's victim is Publora. One REST API call to post to ten social platforms. Ten. LinkedIn, X, Instagram, TikTok, YouTube, Facebook, Threads, Bluesky, Mastodon, AND Telegram — because apparently the MVP was "connect everything humans have ever posted to." | Publora's free tier is a welcome mat for exactly those people — the ones who are long on time-to-value and short on credit cards. | That's no
VC Boom — Leonard wouldn't ship this
An 8-year VC built this. Eight years in venture capital and the best idea he had was... a mail merge with a confidence score. | Workday hits 95% 12-month retention. Salesforce hits 90%. VC Boom doesn't even have a retention number because there's no recurring relationship to retain. | The pricing page says "$297 one-time" with a Spring launch price that "increases July 1." Today is June 9, 2026. T
Honen — Leonard wouldn't ship this
They invented a new category name. And history has a verdict on that move: 90% of tech companies that have gone public over the past five years positioned themselves in EXISTING markets, not new ones. Honen skipped straight to inventing the word "fluflommer." | Cornerstone OnDemand is hideous. It still owns the contract. | Honen is doing all the expensive market education work so that Microsoft Te
Wave — Leonard wouldn't ship this
Wave is 100% free with zero paywall. They have built a structurally identical retention death machine. | If a CSV was too low a bar, what happens when the bar is just... owning a Mac? | A monetization strategy where the customer pays a third party — Groq — and Wave collects exactly zero dollars from that transaction.
Sources from Lenny's Newsletter and Podcast: Lessons from going freemium: a deci
Manus — Leonard wouldn't ship this
Somebody looked at entrepreneurship and said "the hard part is the typing." | Manus cheerfully builds your "premium tinned fish storefront" with Lorem Ipsum descriptions and a hero image of a can of cat food, then tells you it's perfect. | Sample prompts are a band-aid on a broken leg.
Sources from Lenny's Newsletter and Podcast: The Subscription Value Loop: A framework for growing consumer subsc
SellerClaw — Leonard wouldn't ship this
They built a corporate org chart for a dropshipping operation. | "Every action is visible and approvable" doesn't sound like automation, it sounds like a second job. | "Free to start" is a great hook — right up until the agent starts.
Sources from Lenny's Newsletter and Podcast: Why your AI product needs a different development lifecycle (Newsletter), What it feels like when you've found product-
Empromptu AI — Leonard wouldn't ship this
Today's victim is Empromptu AI. They promise to "build your AI app AND train your custom model" — which is the startup equivalent of a restaurant advertising "we cook the food AND wash the dishes." Congratulations. You've described having a kitchen. | Jen Abel says early-stage enterprise ACV sweet spot is $50K to $100K for founder-led startups. But Empromptu is leading with "working features in 10
InsForge — Leonard wouldn't ship this
Today's victim is InsForge — the "agent-native cloud infrastructure platform." They took every backend service a developer needs, bundled it into one platform, and then called it a new market category. Congratulations, you invented the cloud. Again. | Neighborrow had journalists writing love letters, users sending weekly fan emails, and startup competition wins — and zero people actually using the
Vokal — Leonard wouldn't ship this
Congratulations: you built a conference room for robots, and charged humans to sit in it. | Where is the Vokal waitlist full of angry fans demanding a credit card link? | The product page reads like a merger between a pitch deck and a Wikipedia disambiguation page.
Sources from Lenny's Newsletter and Podcast: Positioning (Newsletter), How today's fastest-growing B2B businesses turned their early
Mina — Leonard wouldn't ship this
They've essentially built a coworker nobody hired, who cannot be fired, and who will interrupt your Zoom call to update Salesforce in front of your client. | HeyGen's CEO said it plainly: "Building a cool AI demo doesn't mean we have a product that customers love and is useful." The novelty-driven acquisition leads to the phantom PMF churn cliff — and Mina's entire wow factor is watching an AI spe
Clipto — Leonard wouldn't ship this
Congratulations — you spent how long building this, and the best pitch you landed on is "we're Google Photos, except Google Photos is bad"? | They built a product for the person who buys it and forgot to build a product for the person who approves the budget. | Clipto's positioning literally admits in the first sentence that someone has already done it before. Google did it. Clipto just... took th
Wandesk — Leonard would ship this
No code, no signup, no subscription. They've also decided "no revenue" while they were at it. | Their own website calls it an "App Workshop," which is just a polite word for homework. | At some point "truly yours" becomes "truly free forever and also we are bankrupt."
Sources from Lenny's Newsletter and Podcast: How today's fastest growing B2B businesses found their first ten customers (Newslette
Pancake — Leonard wouldn't ship this
Autonomous agents handling your invoicing, your recruiting, your customer support. The cofounder that never sleeps, never asks for equity, and absolutely will invoice your best client for the wrong amount at 3am. | Tamar Yehoshua watched Glean users ask "what should my top priority be next week" when the system had zero context on their priorities. Now imagine that confusion, but the agent already
Brew — Leonard wouldn't ship this
Brew added a text box and called it a revolution. | They built a four-times-less-effective growth machine and put a gradient on it. | Brew starts at zero, which means they've scientifically optimized for maximum possible churn before a single email gets sent.
Sources from Lenny's Newsletter and Podcast: How to make an impact in your first 90 days (Newsletter), What is good monthly churn (Newslett
Unabyss — Leonard wouldn't ship this
They didn't build a product; they built a buffet where every dish is lukewarm. | Unabyss just volunteered to be the Ask Jeeves of AI context management. | Typeform had one product and still got lost.
Sources from Lenny's Newsletter and Podcast: How to kickstart and scale a consumer business—Step 2: Identify your super-specific who (Newsletter), Ecosystem is the next big growth channel (Newsletter
ModelHub — Leonard would ship this
ModelHub can hit a thousand stars on GitHub and still have no idea which users actually needed it versus which ones just downloaded it because it was four megabytes. | Who does a local LLM developer send their menu bar app to — their models? The blobs in the HuggingFace cache don't have email addresses. | They've built the most developer-friendly tool in the world and then forgot the part where th
Memdex — Leonard wouldn't ship this
Memdex's answer to "give the LLM everything" is "here are your ten most recent conversations, enjoy." | They're not landing and expanding, they're landing and evaporating. | Your product's own pitch explains why the free version is useless.
Sources from Lenny's Newsletter and Podcast: Lessons from going freemium: a decision that broke our business (Newsletter), The Subscription Value Loop: A fram
TestSprite — Leonard wouldn't ship this
50,000 developers onboarded, 100,000 community members — and not a single revenue number anywhere on the site. The $1M ARR benchmark for top B2B companies is 1.5 years from first customer. You know what's conspicuously absent from TestSprite's homepage? The year they signed their first customer. | TestSprite is collecting compliments on the lobby while the hotel is empty. | They send a "fleet of p
Tycoon.us — Leonard wouldn't ship this
Category creation requires a specific enemy to defeat — HubSpot had outbound, Salesforce had spreadsheets. Tycoon's enemy is apparently "having employees," which describes every person on Earth who doesn't want employees. That's not a category, that's a demographic so wide it includes my dentist. | Managing a thousand AI agents in parallel sounds incredible in a demo and sounds like a support tick
StoreClaw — Leonard wouldn't ship this
A startup called Cascade tried that same "horizontal toolkit for a specific buyer" pitch. Their conclusion? "Business analysts at midsize companies no longer faced the diversity of analytical problems they used to, which meant that more opinionated, targeted products could pick up what was left." StoreClaw is building the melting island. | StoreClaw's "connect in minutes" onboarding isn't a moat —
PollyReach — Leonard wouldn't ship this
Today's victim is PollyReach. You tell your AI "book me a table for 7pm" — and it calls the restaurant for you. Congratulations, you have automated the one phone task that already had OpenTable, Resy, and a free app called "your own fingers." | Consumer subscription apps have lower ARPU than B2B SaaS, they have a harder time expanding ARPU than almost any other model, and out of every consumer sub
SocLeads — Leonard wouldn't ship this
SocLeads is competing directly with Meta's own ad targeting and LinkedIn's own Sales Navigator. That's not a partnership, that's poaching. | Google Maps billing you for "any row, which may not include an email" is not a data product. That's a lottery ticket with a monthly subscription fee. | Congratulations, you've automated the part of sales that everyone already hates, and made it dependent on f
Vivago Video Agent — Leonard wouldn't ship this
Forty minutes. For one minute. That's a 40-to-1 suffering ratio. | You're activating a waiting room full of tire-kickers who want to play with a "swarm of AI directors" and will ghost you the second the free trial ends. | That is a churn machine wearing a creative brief as a disguise.
Sources from Lenny's Newsletter and Podcast: Five steps to starting your product-led growth motion (Newsletter),
Loova — Leonard wouldn't ship this
Who looked at video production and thought, "you know what this needs? More steps"? | Loova is out here inventing "AI director" as a category, doing all the hard work of making that term mean something, so that when Meta ships Creator Studio Pro with a director mode, Loova becomes the footnote. | Loova is just a canvas sitting on top of Sora 2, Kling AI, Google Veo — models it doesn't own, didn't
HasData — Leonard wouldn't ship this
HasData's homepage lists Claude, ChatGPT, AI agents, data pipelines, AND CLI users as simultaneous targets. Geoffrey Moore's exact words: "You don't hold a match under a big log." HasData is currently holding one match under six logs. | PayPal dropped every other plan when they went all-in on eBay. HasData's homepage still proudly lists Zillow scrapers, Indeed scrapers, Google Maps scrapers, and e
Spellar AI — Leonard wouldn't ship this
They picked the single demographic most allergic to paying for software and called it a go-to-market strategy. | Spellar has five AI model logos, 20-plus integrations, and zero evidence of a sales motion. The enterprise is going to get eaten, and they'll find out in the renewal email. | Congratulations — you spent how many engineering cycles to build a product whose entire value proposition is "we
Ghost — Leonard wouldn't ship this
Ghost is simultaneously the best host for Minecraft AND Valheim AND Rust AND Palworld AND Counter-Strike. The lesson from a startup that didn't make it is explicit: "Horizontal products win if a specific audience encounters enough use cases that they want one tool to address them all." Gamers don't want one tool for all games — they want the best tool for their game. | Ghost hasn't chosen a bad pa
FlowMarket — Leonard wouldn't ship this
FlowMarket promises a one-minute setup and then hands your company's first impression to an agent that needs constant babysitting. Your brand's opening line to every potential partner is a bot that might be hallucinating your pricing. | They've automated away the one part of sales that actually closes deals: the human being. | Substack is surfacing writers to readers who opted into discovery. Flow
Mindra — Leonard wouldn't ship this
Okay, audit logs are cute — but who actually knows how to use this thing? Airtable had two education problems: users didn't know what the product was for, AND they didn't know how to design a workflow. Mindra has that problem squared: users don't know what AI orchestration is, AND they don't know how to govern a multi-agent team across their entire stack. | Mindra wants you to feel calm handing mi
Radar — Leonard wouldn't ship this
Patrick Campbell — who built a pricing product and should know — said it plainly: "Analytics products are terrible. Willingness to pay for them is terrible; retention for them is terrible; NPS is terrible." Radar is, functionally, an analytics product dressed up in a kubectl costume. | Shaun Clowes — who ran data teams at Atlassian, Salesforce, and Confluent — said data is "more like a compass tha
Jupitrr — Leonard wouldn't ship this
Today's victim is Jupitrr — a company that looked at the entire video production stack, every specialized tool built by teams of experts over a decade, and said: "What if we did all of that, but worse, in one app?" | Atlassian ran the bundled-land experiment and killed it. Their exact conclusion: bundling "really slowed down the product led growth motion." Jupitrr's entire go-to-market is the thin
Plurai — Leonard would ship this
Right, the AI engineer ships it with zero labeled data, zero production history, zero real signal — and Plurai's own FAQ admits they only get "more useful once someone has a business and real data to work with." That's not a product. That's a promise that matures in 18 months. | Plurai thinks "vibe-train in minutes" replaces that entire consultative motion. Good luck explaining intent calibration
SureThing — Leonard wouldn't ship this
You're not paying $30 for a COO. You're paying $30 for a Magic 8-Ball with a LinkedIn profile. | SureThing wants to be a fluflommer before anyone knows what a fluflommer is. | Two refunds totaling $158 is the hero testimonial?
Sources from Lenny's Newsletter and Podcast: How the most successful B2B startups came up with their original idea (Newsletter), Why your AI product needs a different devel
Jet — Leonard wouldn't ship this
Jet has turned a cautionary tale into a product strategy. | Who greenlit "we do everything for everyone" as the positioning? | Viral distribution of a buggy, horizontal agent builder just means your reliability disasters spread faster.
Sources from Lenny's Newsletter and Podcast: Why marketplaces fail (Newsletter)
Evala — Leonard wouldn't ship this
Evala is asking users to care enough to curate a dataset before they've seen a single result. | That's not a self-serve funnel — that's a consulting engagement that forgot to charge. | The product is basically a beauty pageant for AI models — except the judges are your own messy CSV files and the winner is whoever hallucinates least.
Sources from Lenny's Newsletter and Podcast: GTM motions of 30
OpenClaw x Paperclip x Spud — Leonard wouldn't ship this
They didn't build a co-founder. They built an out-of-office reply. | Who is OpenClaw's specific buyer? "Anyone who wants to start or grow a business." That's three billion people and a dream. | A co-founder who promises to validate your idea, run your tasks, build your product, and grow your revenue is not a co-founder, it's a black water pond with a good pitch deck.
Sources from Lenny's Newslett
Beezi AI — Leonard wouldn't ship this
They've invented a brand new category called "AI orchestration platform." You know who else invented a brand new category? Ask Jeeves. Ninety percent of tech companies that have gone public in the past five years were positioned in EXISTING markets — not categories they made up themselves. | Congratulations, you've built a traffic cop for robots, and charged a SaaS subscription for the privilege o
FocuSee — Leonard would ship this
Today's victim is FocuSee — a screen recorder that promises "polished demos in minutes." Seven AI features bolted onto a screen recorder, and their big pitch is... you won't have to edit. Buddy, the editing IS the job. You just sold someone a car and removed the steering wheel. | Mission-critical? Great framing. Here's the mission-critical reality: fewer than 50 consumer subscription apps have EVE
SpeakON — Leonard wouldn't ship this
Thank you for proving my point with better data than I had. Moesta watched Basecamp users scream for Gantt charts for years — and then not a single one left when Basecamp refused to build them. SpeakON's entire TAM might just be people who tweet about hating their keyboard and then keep using their keyboard. | DoorDash went to Tier 2 cities — they didn't ask suburbanites to duct-tape a new steerin
RankAI — Leonard wouldn't ship this
Today's victim is RankAI. "Drop in your website and we handle the rest." Founders, that's not a product pitch — that's what a parking valet says before he loses your car. | Patrick Campbell is very specific about this: products that sit in the middle — not daily-use workflow tools, not fully invisible infrastructure — retain at catastrophic rates. He literally called it "the death zone." RankAI is
Dune — Leonard wouldn't ship this
Geoffrey Moore made a brutal point that's stuck with me: software disruption is "increasingly improbable" because you're not standing on the shoulders of giants, you're standing on the shoulders of people standing on the shoulders of people standing on the shoulders of giants. And Dune isn't even software — it's a plastic puck trying to automate what Command-Tab already does for free. | A tiny sta
Verdent — Leonard would ship this
Verdent promises to "keep working even when you're offline" — it just can't keep users working when they're online. | They've automated the exact problem Replit exists to escape. | That's not a cofounder, that's a ransom note.
Sources from Lenny's Newsletter and Podcast: The rise of the professional vibe coder (a new AI-era job) | Lazar Jovanovic (Professional Vibe Coder) (Podcast), How to make a
Product Hunt — Leonard would ship this
Six dimensions. For a quiz. A quiz that a sticky note and a brutally honest friend could replace for zero dollars. | Product Hunt's quiz is evaluating ALL app ideas across the SAME six dimensions. A gaming app and a B2B SaaS tool are not the same patient, and this quiz is handing them the same prescription. | Product Hunt's core value promise is "find out your idea might be bad." Nobody is going v
Product Description on Product Hunt — Leonard would ship this
A product built for the open web that can't survive a basic HTTP request — outstanding. | PayPal hit a million users by going all-in on exactly ONE wedge: eBay sellers. These folks are simultaneously pitching to three fundamentally different users who have nothing in common operationally. That's not a wedge, that's a pitchfork. | Kevin Weil at OpenAI has 3 million developers on their API precisely
Intent — Leonard wouldn't ship this
Intent's agents will gaslight you into thinking your feature is built and shipped when it's actually a hallucinated mess. | Intent's agents will spend their computational budget trying to understand your codebase instead of building anything useful. | Good luck getting your actual engineering team to adopt Intent when real developers are openly hostile to AI development tools that promise magic bu
AI Co-Builder — Leonard wouldn't ship this
AI Co-Builder promises to build business apps that actually work. They charge you to describe what you need and get a broken app in return — it's like paying a restaurant to cook you food that's still frozen in the middle. | Dan Shipper runs a company with 100% AI-written code and explicitly states that conventional SaaS apps are impossible for non-technical users beyond demos. These people litera
Spotify — Leonard would ship this
Today's victim is Spotify. A music streaming service so desperate to justify its existence, it's literally being used as Google's guinea pig for alternative billing because their margins are too pathetic to afford normal app store fees. | With Spotify's notoriously thin margins paying most revenue to record labels, their contribution margin per customer is so anemic that their customer acquisition
Zoom — Leonard wouldn't ship this
Today's victim is Zoom. "One platform to connect" — because apparently video calls weren't enough, so they decided to add chat, docs, phone systems, contact centers, AI assistants, and probably a coffee machine. They've become the startup equivalent of a Swiss Army knife designed by a committee that never met. | When building horizontal products, "when you try to do something for everyone, you don
Loom — Leonard would ship this
A screen recording tool that somehow convinced millions of people they need to film themselves explaining what a three-sentence email could handle. | Enterprise sales for people who want to avoid talking to other people? That's like selling hearing aids at a death metal concert. | Their homepage brags about "millions of people across 400,000 companies" — which sounds impressive until you realize t
Anthropic — Leonard would ship this
They take forever to ship anything while claiming it's for "safety," which is just slow development wearing a lab coat. | Anthropic takes years between model releases and still can't consistently beat GPT-4. | Meanwhile, they're bragging about helping a Mars rover drive four hundred meters like they invented interplanetary navigation.
Sources from Lenny's Newsletter and Podcast: What is good rete
Superhuman — Leonard wouldn't ship this
Today's victim is Superhuman — the company that looked at Gmail and thought "what this needs is a $30 monthly subscription and the complexity of NASA mission control." | They're literally choosing the churn-maximizing payment structure while calling themselves productivity experts. | Despite claiming "AI that works everywhere," they're still a glorified email client with zero switching costs.
Sou
Figma — Leonard would ship this
They've built a design tool that's so collaborative, your entire design team can watch you screw up in real time. Nothing says "healthy workplace culture" like live-editing someone else's mockup while they're still working on it. | Welcome to the plateau, Figma. | Figma's building a beautiful house on a foundation of quicksand.
Sources from Lenny's Newsletter and Podcast: Scaling your B2B growth
Duolingo — Leonard would ship this
They promise "free, fun, effective" learning, but their mascot literally threatens you with death memes when you skip a lesson. | They had to steal tricks from FarmVille to make people care about conjugating French verbs — that's not product-market fit, that's digital manipulation. | When your growth engine breaks, you're left with an owl making threats to teenagers.
Sources from Lenny's Newslett
Linear — Leonard would ship this
They're billing themselves as purpose-built for modern teams, which is startup speak for "we couldn't compete with Jira so we added chatbots." | Linear's expensive AI-powered features risk becoming another underutilized enterprise tool that developers ignore while they go back to their trusted workflows. | You cannot cross the chasm of product-led growth because enterprise buyers need serious hand
Slack — Leonard would ship this
They turned workplace chat into a notification casino where every ping is a slot machine pull. Now they want to add Slackbot AI to coordinate your work — because what busy professionals really need is a robot scheduling their existential dread. | They're charging enterprises for AI that can't think while teams average 43 apps already. Forty-three! They built a digital filing cabinet with a chatbot
Notion — Leonard wouldn't ship this
Today's victim is Notion — the AI workspace that works for you, as long as "working" means replacing forty-seven different tools with one tool that does all forty-seven things poorly. | Notion brags about penetrating 62% of Fortune 100 but conspicuously avoids mentioning their own retention rates. | Their pricing calculator shows $4,080 in annual savings, implying high price points — but they stil
Recommended
15 Minute Mysteries: The Deep Dive
15 minutes de grâce et de vérité
15 Minutes of Infamy
15 Minutes with Jesus: Christian Meditation, Guided Prayer, Bible Study, Emotional Healing, Devotional, Hear God’s Voice
180Podcast.
1856 Podcast-YMCA of South Hampton Roads
1984
1984, by George Orwell
19 Keys Presents High Level Conversations
19 Observations on mining and refining of critical minerals
1A
1Dime Radio
