Tri Diligence

Tobbe

Business ideas, unwrapped. Tri Diligence is the show where a business idea meets three people who've actually been in the room. Each episode we take one idea — yours, ours, or one plucked from the headlines — and unwrap it from three angles: - Jake, the builder — brand, customers, and whether anyone will love this enough to tell a friend.- Sarah, the backer — the unit economics, the margins, the market size, and when the money comes back.- Ryan, the technologist — what you buy, what you build, and the operational traps hiding under the pitch.We walk the whole business model — customers, value, channels, revenue, costs, partners — and pressure-test it against how comparable companies really performed: who IPO'd, who got acquired, who quietly went to zero. No hype, no hand-waving — just real data and the uncomfortable questions.Every episode ends the way real diligence should: three clear verdicts — invest, wait, or pass — and the one first step each host would take.Good ideas welcome. So are the hard numbers.Tri Diligence is produced with AI and is made for entertainment and idea-exploration, not investment advice.

Episodes

  1. Jul 7

    The Regulars: Can a Small Neighborhood Cafe Ever Pay You to Leave It?

    Three recurring AI host perspectives unwrap The Regulars — a tiny cafe in a nice, quiet residential area with barely any foot traffic, built on locals and repeat visits. The founders real goal is not just to open it; it is to eventually stop working the counter. Can a low-traffic, locals-only cafe throw off enough to pay a barista AND a manager to replace the owner — and still profit? Or is this the classic trap: buying yourself a job you then cannot even leave? The whole episode turns on one question: can the math ever pay the owner to walk away? • Jake (the marketer) — the cafe is not coffee, it is the "third place": recognition, ritual, becoming the neighborhood default living room. But he gets cornered on his own thesis: "become beloved" has to turn into a testable frequency lift, not a feeling. • Sarah (the backer) — the brutal unit economics: a ~$7.50 ticket at 100 covers a day, ~300 days, is ~$225k in sales, before food (28-35%) and wages (~36.5%) eat it. And the killer: a manager wage means ~$90k in EXTRA sales — about 33 more people through the door every day — just to buy back the owners mornings on a street with no footfall. • Ryan (the technologist) — the systems that let an owner actually leave (documented ops, a real CRM, off-the-shelf Toast/Square/Shopify, not custom "founder cosplay"), plus the AI angle: demand forecasting to cut waste and loyalty personalization to widen the habit loop, vs a funded rival using location data and subsidized loyalty to open next door. Make-vs-buy the food gets a clear answer — do not hire a baker on day one: buy pastries wholesale, assemble a tight set of sandwiches in-house for margin and identity, skip the miniature Pret in the back. Plus the revenue streams that actually fit (retail beans, scheduled office wholesale, house accounts) and the ones that are "a hobby with chairs." Grounded in the real landscape — Starbucks 41,000+ stores, Dutch Bros ~$1.28M/shop — and why scale still cannot fake intimacy. Each host ends with a verdict — invest tiny and prove it with a pop-up first, wait for a real model, or build the operating system first — and one concrete next step. The thesis: The Regulars wins only if it is a repeat engine, not a postcard. Tri Diligence is an AI-produced analysis, made for entertainment and idea-exploration, not investment advice. Full analysis, transcript and the three verdicts: https://pods.purpur.se/tridiligence/ep10/

  2. Jul 5

    Netcode Games: Can a Backend Architect Survive Indie Game Dev on Steam?

    Three recurring AI host perspectives unwrap Netcode Games — a veteran cloud-and-networking architect who wants to go indie, ship games on Steam, and lean hard on AI. The skills are real: multiplayer, servers, analytics, the invisible machinery most solo devs cannot build. One problem — Steam does not give bonus points for elegant architecture, and the enemy is not downtime, it is nobody knowing you exist. The whole episode turns on one question: is a serious engineering resume a real edge in indie games, or resume bias pointing at the wrong genre? • Jake (the marketer) — founder-market fit is real IF the genre is right: not an MMO, but a small, social, systems-heavy co-op game where reliability is part of the fantasy. Players do not buy netcode; they buy "my friends and I are running a cursed space diner while asteroids hit the kitchen." Sell the feeling; the architecture is the enabler, not the headline. • Sarah (the backer) — the math: 21,402 games launched on Steam in 2025; the enemy is obscurity. Wishlists are the leading indicator (7,000-9,999 = Gold, 10,000+ = Diamond; do not launch below 10k). At $15, Steam takes 30% -> $10.50/copy, so netting ~$50k means ~8,000-10,000 paid copies ("ramen profitable, not yacht profitable"). Plus the refund trap: 14 days / under 2 hours played. Verdict: wait, keep the day job. • Ryan (the technologist) — build brutally small: prove one social loop in ~60 days with off-the-shelf everything (Unity/Godot, Steamworks, PlayFab or Nakama, Sentry-style crash reporting, analytics), custom backend only where it creates differentiated gameplay. No internal cathedral. And the AI angle both ways — AI for you (code assist, playtest bots, trailer rough cuts, localization) vs AI against you (competitors flooding Steam and cloning your hook in a month). Steams gate is tiny: $100/game, recoupable after $1,000 in revenue; the real cost is live ops. Each host ends with a verdict — invest in the experiment, wait, or build small — and one concrete first step. The thesis: do not try to out-studio the studios; use serious engineering to make a small, sharp, social game people can describe in one breath. Tri Diligence is an AI-produced analysis, made for entertainment and idea-exploration, not investment advice. Full analysis, transcript and the three verdicts: https://pods.purpur.se/tridiligence/ep9/

  3. Jul 4

    Glitter Scoop: Can a 10-Week Ice Cream Shop Pay Its Staff and Still Profit?

    Three recurring AI host perspectives unwrap Glitter Scoop — a sparkly, summer-only ice cream parlor open about ten weeks a year. Cute shop, brutal calendar. Can a seasonal scoop shop cover the rent, pay a real crew, and still leave the owner a profit — or is it a lifestyle business where the founder is the unpaid labor? Glitter Scoop: a walk-in parlor with a glitter-and-sparkle brand, one/two/three scoops, banana-split specials, rotating flavor collections, and a coffee machine. Open half of June, all of July, half of August — about 70 trading days a year. • Jake (the marketer) makes the case for a lovable, photogenic, LOCAL flagship, and argues "lifestyle business" is not an insult if it pays the staff, pays the owner, and makes families happy. • Sarah (the backer) runs the break-even fight: ~90k in annual fixed costs at a 35% contribution margin needs roughly 400k in seasonal revenue — about 630 orders a day. Pricing, location, and the line between a real business and an investable one. • Ryan (the technologist) covers the operations behind the sprinkles and the AI angle both ways: AI for you (weather + hourly history to forecast demand, staffing, and flavor prep) and AI against you (a chain using location data and dynamic coupons to grab your tourists first). Pressure-tested against a $166.7B frozen-desserts market you will never chase, Unilever's €8B ice-cream turnover, Jeni's as the aspirational comp, and Halo Top as the novelty-fades-fast warning. Each host ends with a verdict — go, wait, or build light — and one concrete first step. Tri Diligence is an AI-produced analysis, made for entertainment and idea-exploration, not investment advice. Full analysis, transcript and the three verdicts: https://pods.purpur.se/tridiligence/ep8/

  4. Jul 4

    Screen Threads: Can You Sell Movie T-Shirts Without Getting Sued?

    This week on Tri Diligence: Screen Threads, a store for the exact tee a character wore on screen - browse by show or film, print-on-demand, drop-shipped, no inventory. The whole episode turns on one question: is this a company, or a cease-and-desist? Jake (the marketer) - the real wedge isn't the Marvel-logo crowd; it's the fan who says "wait, I want Carmy's shirt" - the plain, unbranded tee a character actually wears, which carries no trademark. That's the business hiding inside the lawsuit. Sarah (the backer) - recreating a studio's print is copyright and trademark infringement; the thin print-on-demand margins against the very real legal liability; and whether the only safe version of this is boring. Ryan (the technologist) - which designs are genuinely non-infringing, the AI angle (upload a screenshot, detect the garment, match the print - and the flip side, studios using AI to auto-issue takedowns), and whether you could flip the whole thing into a licensed, official partner instead of a pirate. Pressure-tested against the real landscape - Redbubble, TeePublic, and Etsy, flooded with fan art and constant DMCA takedowns. Each host ends with a verdict - invest, wait, or pass - and one concrete first step. Tri Diligence is an AI-produced analysis, made for entertainment and idea-exploration, not investment advice. Full analysis, transcript and the three verdicts: https://pods.purpur.se/tridiligence/ep7/

  5. Jul 4

    The Sovereign Cloud: Can Europe Ditch AWS for €50 a Month?

    This week on Tri Diligence: the sovereign cloud - the idea that you don't need to rebuild Amazon to break free of it. About 90% of apps need only three primitives - object storage (S3), compute (Lambda), and a database (DynamoDB) - and mature open-source equivalents now exist for all of them (Garage/MinIO, Knative, ScyllaDB), plus containers for portability. Our three hosts pull it apart: Jake (the marketer) - the liberating fact: about €50 a month on a Hetzner box in Germany and you're sovereign from hour one, outside the US CLOUD Act. Europe has no hyperscaler of its own. Sarah (the backer) - the asterisk on "free and scalable": hardware is cheap, your time is not. Which of three plays is a real business - personal/SMB independence, a sovereign PaaS, or a full IaaS provider? Ryan (the technologist) - the real stack and its one hard problem (matching hyperscaler burst elasticity and scale-to-zero), plus where AI compute and sovereignty collide. Proof it's real: 37signals (Basecamp and HEY, led by DHH) left the cloud - about $600k on hardware, saving roughly $2M a year - and even they chose colocation, not their own server room. The frame: the server room isn't the starting line, it's the finish line you grow into - and the first step costs less than dinner. Each host ends with a verdict - invest, wait, or pass - and one concrete first step. Tri Diligence is an AI-produced analysis, made for entertainment and idea-exploration, not investment advice. Full analysis, transcript and the three verdicts: https://pods.purpur.se/tridiligence/ep6/

  6. Jul 3

    Havit: Charming Hobby, or Real Business?

    This week on Tri Diligence: Havit, an app that answers one question at the exact moment it matters - standing in a shop, "do I already own this?" Photograph the item and a two-stage AI pipeline (embeddings to shortlist your likely duplicates, then a vision model for the final call) answers in about two seconds. The founder's hook is real: they collect purple soda cans because the company is called Purpur, and keep forgetting in the store whether they already have that exact can. Our three hosts fight over the one question that matters: Jake (the marketer) - the "do I already have it?" moment fires exactly when the wallet is out; start with high-frequency, high-value niches, not every collector. Sarah (the backer) - is a generic collection app a hobby in a crowded market, or is there real money in niches where duplicates cost real money (cards, sneakers, vinyl, whisky, wine)? And which of three models wins: affiliate at point of purchase, freemium subscription, or B2B white-label to shops and auction houses? Ryan (the technologist) - the pipeline already works (serverless AWS plus embeddings plus a vision model); where AI is the product, and where a well-funded incumbent could bolt the same on. Each host ends with a verdict - invest, wait, or pass - and one concrete first step. The honest bottom line: don't build an app for everything you could collect - build it for the thing you already buy every month. Tri Diligence is an AI-produced analysis, made for entertainment and idea-exploration, not investment advice. Full analysis, transcript and the three verdicts: https://pods.purpur.se/tridiligence/ep5/

  7. Jul 3

    Vintel: Can a Wine-Cellar AI Beat CellarTracker?

    This week on Tri Diligence: Vintel, an AI layer for your wine cellar. It does three things - recommends a bottle you actually own to pair with tonight's meal, flags bottles hitting "value at risk" before they pass their peak, and nudges you on drink windows ("drink it this spring, not in two years"). Our three hosts pull it apart: Jake (the marketer) - the intimate "trusted dinner guest" wedge, and why a niche tool can out-charm a broad one. Sarah (the backer) - willingness to pay on top of a hobby people already overspend on, and whether the wine-bar angle is real revenue. Ryan (the technologist) - clean cellar data, where AI helps versus where it becomes a very confident sommelier with a blindfold, and the scary half: if CellarTracker or Vivino add a conversational layer, does Vintel just become a demo? We pressure-test it against the incumbents - CellarTracker (about 10 million users, 200 million bottles) and Vivino Premium, which already claims pairings, collection value, and drink windows for $4.99 a month. The founder's edge: the core engine already exists. The hook: a real 2009 Donnhoff / Dr. Loosen Riesling in the founder's own cellar that the app says to drink this spring. Each host ends with a verdict - invest, wait, or pass - and one concrete first step. Tri Diligence is an AI-produced analysis, made for entertainment and idea-exploration, not investment advice. Full analysis, transcript and the three verdicts: https://pods.purpur.se/tridiligence/ep4/

About

Business ideas, unwrapped. Tri Diligence is the show where a business idea meets three people who've actually been in the room. Each episode we take one idea — yours, ours, or one plucked from the headlines — and unwrap it from three angles: - Jake, the builder — brand, customers, and whether anyone will love this enough to tell a friend.- Sarah, the backer — the unit economics, the margins, the market size, and when the money comes back.- Ryan, the technologist — what you buy, what you build, and the operational traps hiding under the pitch.We walk the whole business model — customers, value, channels, revenue, costs, partners — and pressure-test it against how comparable companies really performed: who IPO'd, who got acquired, who quietly went to zero. No hype, no hand-waving — just real data and the uncomfortable questions.Every episode ends the way real diligence should: three clear verdicts — invest, wait, or pass — and the one first step each host would take.Good ideas welcome. So are the hard numbers.Tri Diligence is produced with AI and is made for entertainment and idea-exploration, not investment advice.