We talk about why every can is lab-tested, hiring slowly and working with a co-packer, going from 100 views to viral videos that actually sell, paid ads and subscriptions at a 0.5 ROAS, net 30, MOQs and forecasting inventory, why Amryth is skipping retail for now, building a 150-creator TikTok Shop tribe, the road to $1M and winning back customers, and why copycats don’t make it zero-sum.

Watch on YouTube:

. More from me at benji.app.

Check out Amryth at drinkamryth.com.

Timestamps

• (00:00) – Introduction

• (00:28) – What Is Mad Honey?

• (01:20) – From Finance Intern to Alcohol Alternatives

• (04:52) – From Nectar to Amryth: Building a Premium Brand

• (05:44) – Raising the Price to $40 and Selling More

• (08:52) – Pitch Competitions and a Supplier Faking It With THC

• (10:05) – Going to the Source: A 5 A.M. Call and a Flight to Nepal

• (12:46) – Honey Hunters, Giant Bees and Food Poisoning

• (17:08) – Why Every Can Is Lab-Tested

• (18:13) – Hiring Slowly and Working With a Co-Packer

• (21:31) – From 100 Views to Viral Videos That Sell

• (23:42) – Paid Ads, Subscriptions and Scaling at 0.5 ROAS

• (27:07) – Net 30, MOQs and Forecasting Inventory

• (29:13) – Smaller Cans and the Reverse Tolerance

• (31:08) – Why Amryth Is Skipping Retail (For Now)

• (36:24) – Cutting the BS: His Day and His Lean Team

• (40:38) – Starting With Under $5K and Credit Card Float

• (44:46) – Building a 150-Creator TikTok Shop Tribe

• (48:53) – The Road to $1M and Winning Back Customers

• (53:50) – Copycats, Competitors and Why It’s Not Zero-Sum

We talk about the Ponder thesis (agents that orchestrate AI models instead of one more chat box), why not talking to customers early was his biggest artist-founder mistake, how new models finally got taste, why the bottleneck is product form factor and not models, product-led growth vs. “distribution is the moat” when incumbents can spend $1M a week on ads, building desktop first, Claude Code and DaVinci MCPs, and why the incumbents in video editing should be scared. I also try Ponder live and get a podcast edit in 3 prompts.

Watch on YouTube:

. More from me at benji.app.

Check out Ponder at ponder.ai, and they’re hiring at ponder.ai/careers.

Timestamps

• (00:00) – Introduction: Building the World’s Best AI Video Editor

• (01:51) – Ponder’s Focus: Winning the “Lovable Moment” for Video

• (02:40) – Origin Story: NYU Film School and Joma Tech’s First Intern

• (06:15) – First Company: WLab and Virtual Production

• (11:01) – Selling WLab at 21 and Naming Ponder

• (12:26) – Going Back to School: Runway Gen-2 and NYU ITP

• (15:51) – The Ponder Thesis: Agents That Orchestrate AI Models

• (17:38) – Artist Founder: Why Not Talking to Customers Was a Mistake

• (19:53) – Beyond Chat: The Future Interface for Creative AI

• (22:04) – AI Got Tasteful: How New Models Changed Video Editing

• (23:42) – The Real Bottleneck: Product Form Factor, Not Models

• (26:49) – Go-to-Market: Product-Led Growth vs. “Distribution Is the Moat”

• (29:55) – Hiring Philosophy: People Who Really Want to Win

• (31:24) – Ponder Studios: Starting With Creators

• (34:20) – Why Desktop First, Not Mobile

• (36:48) – Organic Growth: Half of Users Come From Invites

• (38:21) – Claude Code, DaVinci MCPs, and Setup Friction

• (41:04) – Who Should Be Scared? Why the Incumbents Are Cooked

• (43:42) – Benji Tries Ponder: A Podcast Edit in 3 Prompts

• (45:21) – Everyone Has a Story Worth Telling

• (47:49) – Monetization and Unit Economics

• (48:49) – Leadership Style: Managing a Team of “Cannons”

We talk about CAC, LTV and bundles, getting 30-day credit from a factory with no contract and cutting unit costs in half, hitting the ads ceiling and moving into supplements, the offers and subscriptions that put him $50K down, why product, offer and ads are the only three levers, hiring in Southeast Asia, what happens if Meta bans your ad account, and the part nobody talks about: losing the drive after making it, then selling the company and feeling lost.

Watch on YouTube:

. More from me at benji.app.

Check out Shen on Instagram at instagram.com/weeyishen.

Timestamps

• (00:00) – Introduction

• (00:32) – Meet Shen

• (02:08) – Midnight Product, 6 A.M. Store, 10 A.M. Sale

• (03:42) – Pre-Selling $100K Before the Product Existed

• (04:50) – Sourcing From Chinese Factories With Zero Experience

• (08:18) – Shipping With a 3PL and Scaling Purchase Orders

• (11:17) – Borrowing $10K Every Weekend to Run Ads

• (12:01) – No Face, Just Meta Ads: CAC, LTV and Bundles

• (14:34) – Going Solo and Losing 8 Kilos

• (15:55) – 30-Day Credit, No Contract and Cutting Unit Costs in Half

• (21:38) – Hitting the Ads Ceiling and Moving to Supplements

• (25:26) – Offers, Subscriptions and Going $50K Down

• (29:55) – Why He Never Raised VC

• (32:12) – 1.5M With No Books or Accounting

• (33:51) – Product, Offer, Ads: The Only Three Levers

• (36:52) – Hiring in Southeast Asia and the Fight for A-Players

• (40:47) – What Happens If Meta Bans Your Ad Account

• (42:26) – Losing the Drive After Making It

• (45:24) – Selling the Company and Feeling Lost

We talk about whether every model upgrade makes Imprezia stronger or weaker, why he does every job himself before hiring (you can’t lead a 2,400 Elo team at 800 Elo), solving cold start in a two-sided marketplace, why cold email doesn’t work anymore, why search is dying and where ad budgets go next, whether Google ends up owning AI advertising, how a real-time auction actually prices an ad and sets CPMs, why context makes ads worth more, hiring hunters instead of farmers, and whether AI kills the big ad agencies.

Watch on YouTube:

. More from me at benji.app.

Check out Imprezia at imprezia.com.

Timestamps

• (00:00) – Intro

• (01:00) – Building Google AdSense for AI Apps

• (01:34) – Why Ads? Joining Amazon Ads in 2019

• (03:11) – From Meta GenAI Ads to Starting His Own Company

• (04:39) – How Fast AI Models Are Moving

• (06:21) – Does Every Model Upgrade Make the Business Stronger or Weaker?

• (08:02) – Do Everything Yourself Before You Hire

• (09:42) – You Can’t Lead a 2,400 Elo Team at 800 Elo

• (11:42) – Solving the Cold Start Problem in a Two-Sided Marketplace

• (12:23) – Go-to-Market: Why Cold Email Doesn’t Work Anymore

• (14:10) – Business Model: Revenue Share, No Retainer

• (14:50) – Why Ad Tech Can’t Be Fully Self-Serve

• (15:45) – The Future of Marketing in an AI World

• (17:50) – Every Billion-User Product Runs on Ads

• (18:14) – Search Is Dying: Where Ad Budgets Go Next

• (19:37) – Will Google Own AI Advertising?

• (21:33) – How an Ad Gets Priced: The Real-Time Auction

• (23:31) – Why Context Makes Ads More Valuable

• (24:40) – How CPMs Are Set

• (25:41) – The Data Flywheel Behind Better Ads

• (27:22) – Revenue Share and Where the Network Performs Best

• (30:44) – Hiring: Hunters, Not Farmers

• (32:44) – KPIs: Total Transparency Over Individual Metrics

• (34:56) – The Biggest Bottleneck: Slow-Moving Advertisers

• (38:08) – Will AI Kill the Big Ad Agencies?

We talk about his playbook (find an outcome people already pay for that has no app, before any keyword tool can see it), why he chose a lifestyle business over venture (time for gene editing and biology), why being a solo founder is mentally hard, what to do differently at 21 vs. 45, and why the next wave is personal agents that actually get things done for you.

Watch on YouTube. More from me at benji.app.

Check out Ali's apps at renaissanceapps.com.

Timestamps

(00:00) – Intro: Renaissance Apps, 60+ apps, all solo
(00:33) – Cuff: building a dating app right before COVID
(02:41) – Why the future of dating runs through AI agents
(05:43) – Dead for six months, then several million downloads
(07:30) – Zero CAC, and buying users to balance a marketplace
(08:55) – Moats: hard to build, network effects, App Store approval
(11:03) – Study tool apps and the ChatGPT arbitrage window
(12:11) – The playbook: real demand that no keyword research shows
(15:19) – Lifestyle business vs. venture scale
(17:35) – 63 apps, 95% failures: the probability game
(19:15) – The mental cost of going solo
(21:19) – Why he chose a lifestyle business
(23:57) – Quitting Meta with six months of runway
(27:09) – Advice at 45 vs. at 21
(28:40) – How to win when everyone has the same AI tools
(32:56) – Selling a study app for multi six figures
(36:21) – What's next: agents that get things done
(43:27) – The future of work: managing agents
(48:10) – One mega markdown file: SOPs for prompting

We talk about why running ads too early was "the worst decision" of his life, why everything at Omi is open source, his hiring formula (your top achievement divided by your age), giving 80% of the equity to the team, one metric per person, six-day weeks, and what he looks for in a partner.

Watch on YouTube. More from me at benji.app.

Follow Nik on X (@kodjima33). Omi is hiring: omi.me/careers.

Timestamps

(00:00) – Introduction: from $2M to $7M ARR in two weeks
(01:20) – Origin story: growing up in Russia
(04:01) – First company: a crypto trading platform
(05:21) – Second company: a freelance marketplace
(07:28) – The painful lesson: why running ads too early failed
(10:02) – Starting Omi: from hackathon to product
(12:59) – Open source philosophy and competitive advantage
(17:02) – Hardware journey: from DIY to manufacturing
(18:22) – Organic growth: Twitter and affiliates
(22:47) – The paid ads breakthrough: 3x growth in two weeks
(23:57) – Hiring philosophy: finding great talent
(32:45) – Equity structure: an 80% ESOP
(38:39) – Leadership style: one metric per person
(45:41) – Work-life balance and personal philosophy
(56:23) – Making friends and building relationships
(1:00:52) – Dating: learning as a core criterion
(1:04:03) – Asking AI about girlfriends, wives and kids

We talk about finding a technical co-founder before AI could write code, the $8K influencer post that did nothing, remote vs. in-person teams, how he'd launch an app today, and the new feature that did half of last month's installs in a single day.

Watch on YouTube. More from me at benji.app.

Follow Luis on X (@luisbnyc) and Instagram (@luisbarreraxyz).

Timestamps

(00:00) – Intro
(00:23) – Meet Luis: Split, Sway AI and Sona
(00:35) – Split: learning to code and finding a co-founder
(02:30) – Split's growth: word of mouth to 500K users
(09:19) – Sway AI: an idea found on Reddit
(16:14) – Going viral, the $30K ceiling and the influencer burn
(24:02) – $0 to $2M ARR in six months with ads
(26:11) – Sona: an astrology app with his sister
(32:25) – Choosing co-founders, remote vs in person
(36:01) – How he'd launch an app today
(40:30) – Eight apps in: why he keeps building
(44:59) – His sister as creative director
(50:53) – How Sona stays accurate and spreads by link
(58:13) – Astrocartography: half a month's installs in one day

I got into the consumer app space because I witnessed an app called Umax make millions. Umax is an API wrapper that rates how attractive you are and people had to pay a subscription fee to gain access to the result.

This app made me realize how much demand was sitting there, unclaimed, in the consumer app space. Building a product that solves a random problem for people and getting paid for it sounded too good to be true. I decided to officially double down on consumer apps.

Keep in mind this was in the pre-claude code era where code was still relatively expensive. A month later, I shipped Pillar, a self-improvement tracker.

Within three weeks I scaled it to 50M+ organic views and 8K downloads, and I hadn't spent a dollar on ads. That's about as clear a signal of product-market fit as you get. Even until this day we still have new users paying us.

Then I made the mistake that first-time founders make. I lost focus.

I stopped doing the boring work as obvious bugs piled up and started thinking about the next idea. In retrospect I should have stayed focused on one singular product and made it good. A few apps in the same category launched after I'd already moved on went on to do hundreds of thousands in profit a month. The market hadn't changed. The timing hadn't shifted. I simply just left the market because I was not "interested" anymore.

The same thing happened with my second app called HeightGPT. It is a GPT wrapper that claimed it could predict and maximize one's height. A product that was built for social media virality. The plan did work out and 67K people downloaded it.

Fun fact here is that I never used it personally for more than 5 minutes a day. I never talked to a single user and shipped almost nothing for months. 67K people downloaded my app and got silence back.

After voluntarily abandoning two products that clearly had traction, I went on to build more products that ended up in similar situations.

Now I'm working on Snag, which shows you free stuff near you. It's simple information arbitrage: someone's getting rid of something, someone else wants it, we connect them. We just crossed $30K MRR recently.

The difference this time is that I actually took my time and think from a user's perspective. I open Snag on weekends. Bits of my apartment came from it. One Sunday I watched a free mechanical keyboard appear a mile from me so I went and got it. Best feeling ever. When you use your own product, you don't need systems to make yourself keep working on it. The small annoyances bother you personally, so you fix them.

I think the most underrated question in startups is really "do you use your own product?" It sounds too obvious to be useful, but it isn't. When I look at the products that compound, they're almost always built by founders who live inside their own creation.

I still wonder what Pillar would look like if I'd stayed six more months. I still think about those 67K HeightGPT users sometimes.

One realization after burning months on failed projects and scattered ideas is that I wasted far too much time trying to figure everything out by myself.

Every problem I kept slamming into had already been solved dozens of times by someone else. I just refused to look because I convinced myself I needed to discover it my own way. That stubbornness was arrogance dressed up as independence.

I fell into what I now call originality obsession. I treated any form of borrowing as weakness or even shame. I'd rather crash into every wall in the name of pure invention than study how someone who'd already succeeded built a smoother path. It felt noble on the surface. In reality it was slow, expensive, and mostly just low-level repetition of other people's past failures.

The proof was always right in front of me. Ali rewatched his idols' fights frame by frame and drilled their moves until they became his own, while Spielberg and Lucas still trade scripts and storyboards and openly borrow each other's rhythm. The people at the very top of their fields are constantly copying, calibrating, and stealing from one another, and I was pretending I could outrun all of that on sheer originality alone, which is really just bad math dressed up as conviction.

Munger taught this quieter than anyone. He never set out to invent grand new theories. He pulled classic models from physics, biology, psychology and applied them ruthlessly to business, investing, and life. The real shift happened when I stopped romanticizing originality and started copying deliberately. I began seeking out people who had already reached the outcomes I wanted and treated them as living maps. I closed my ego long enough to do a brutal disassembly.

That single change, copy first then iterate, has saved me more time and pain than almost anything else.

All the greatness I admire started with humble imitation. The breakthroughs came later, only after a deep, honest possession of someone else's proven logic.

The most practical reminder I can give myself every morning: stop trying to be the first to everything. Start by becoming a good student of those who already figured it out. The rest tends to follow quietly from there.