Pat Walls opens with a cold-start hook that flips conventional startup wisdom on its head. Most founders fail because they can't get users; Eyal and Yali had users — 20 downloads a day right out of the gate — but almost none of them stayed after the free trial ended. ^{[1] — Eyal "Stuck at $2K MRR despite downloads: Despite strong download numbers, PropGPT could not push past $1,000–$2,000 MRR due to poor product rete…" 00:34 Play} That mismatch trapped them at $1,000–$2,000 MRR with no obvious way out. The intro frames the episode's core argument: distribution is necessary but not sufficient. If the product fails the user, no amount of marketing will save it. Pat teases the resolution — a complete rebuild, a 10-week sprint to $30K MRR — and introduces the two college founders who lived through it.

With a brief but confident introduction, Eyal and Yali lay out PropGPT's current state: $30,000 a month, 40,000+ downloads, 3,000+ paying customers, a 48% conversion-to-trial rate, and $3.30 revenue per download. ^{[1] — Eyal "Over 40,000 downloads total: PropGPT has accumulated over 40,000 total downloads since launch." 02:23 Play} The product uses machine learning to analyse sports betting picks, giving users pre-screened bets rather than raw data. But these numbers didn't come easily. The founders quickly acknowledge they got stuck at $1,000–$2,000 MRR for months before the breakthrough. Pat presses them to go back to the beginning — before the App Store launch — to understand what the first version actually looked like and why it didn't work.

Building the first version of PropGPT was no small feat — it took about five months, partly because ChatGPT was still new technology at the time. Eyal describes launching in the middle of the NFL season, riding the wave of sports fan engagement, and initially feeling like the momentum was real. But beneath the surface, the metrics were already sending warning signals. Users were downloading, some were starting trials, but the retention and paid conversion numbers didn't match the excitement. The founders were caught in the classic trap: confusing activity for traction, and mistaking downloads for validation.

Pat Walls pauses the interview to deliver a sponsored message for Starter Story's Black Friday deal — described as one of the best deals the platform has ever offered. He references last year's deal selling out in just two hours as social proof of demand and urgency. Viewers and listeners are directed to starterstory.com/blackfriday or the first link in the description to register for early access. The message is brief, direct, and leans heavily on scarcity and past performance to drive action before returning to the main interview.

Eyal walks through PropGPT's initial go-to-market strategy: influencer marketing, informed by his experience watching other successful app founders use it effectively. They spent a few thousand dollars and got consistent results — around 20 downloads a day, converting to 5 to 10 trial users daily. ^{[1] — Eyal "20 downloads/day at launch: PropGPT averaged 20 downloads per day right after launching on the App Store through influencer marketing." 00:06 Play} For a while, it felt like forward motion. But the MRR number wouldn't budge past $1,000–$2,000, no matter how they dialled up the spend. That plateau was the moment of clarity — something was critically wrong with the app itself. The distribution engine was working; the product simply wasn't delivering enough value to turn trial users into paying subscribers.

The turning point arrived when Eyal and Yali stopped trying to market their way out of a product problem and went back to basics. They shut down all marketing, went into what Pat calls 'the cave,' and spent four months rebuilding PropGPT from the ground up. ^{[1] — Eyal "Users didn't want a sports betting analytics tool — they wanted to be told the answer. Eyal realized their app was making users do the work…" 03:58 Play} The central product insight was deceptively simple: users weren't coming to the app to do their own sports betting analysis. They wanted the answer handed to them — the equivalent of being given the test answers rather than working through the questions themselves. The original PropGPT made users navigate sportsbooks and check their own picks, when all they really wanted was a curated list of pre-analysed bets. That single realisation drove the entire redesign — and would prove to be worth every week of lost marketing time.

Drawing a direct line from Eyal and Yali's success to the power of knowing how to build with AI, Pat promotes Starter Story Build — a live, cohort-based accelerator designed to take founders from idea to working product in just a few weeks. The pitch is grounded in the episode's narrative: Eyal and Yali were able to iterate and rebuild because they had the technical skills to do so. Starter Story Build promises to give non-technical founders those same capabilities through AI tooling. The call-to-action directs viewers to the first link in the description to secure a spot before the next cohort begins.

With the rebuild behind them, Eyal and Yali share the structured playbook they wish they'd had from day one. Step one is knowing exactly who you're building for and what their pain points are. Step two is becoming data-obsessed — not just looking at top-line numbers but understanding what the conversion funnel is telling you at every stage. ^{[1] — Eyal "Step 1: know exactly who you're building for. Step 2: worship your data. Step 3: obsess over in-app analytics to find drop-off points. Step…" 06:54 Play} Step three, Yali adds, is obsessing over in-app analytics: tracking feature clicks to identify your most compelling value proposition, and watching onboarding screens to see exactly where users drop off. Step four is scaling through influencer marketing — but only after the product is ready to convert. The evidence is vivid: their 70th influencer video hit 600,000 views and pushed ARR from $8,000 to $38,000 in three days. ^{[2] — Eyal "Their 70th influencer video hit 600,000 views and single-handedly pushed PropGPT's ARR from $8,000 to $38,000 in three days. Influencer mar…" 08:07 Play} One video, three days, nearly five times the revenue — the power of distribution applied to a product that finally worked.

Pat asks for transparency on the numbers — and Eyal delivers. ^{[1] — Eyal "~50% profit margins: PropGPT runs at roughly 50% profit margins after accounting for marketing, data APIs, hosting, and tooling costs." 09:15 Play} PropGPT is built on a React Native codebase with TypeScript and Python powering the machine learning algorithms and automated data fetching. The database runs on Neon at $10/month, RevenueCat handles subscription management at 1% of revenue, and Superwall manages the paywall at $0.20 per conversion. Real-time sports data APIs cost roughly $100/month, and LLM costs — despite the AI-heavy product — are just $20/month and falling. The biggest cost by far is marketing at $10,000/month, predominantly influencer campaigns. After all that, margins sit at approximately 50%, a strong return for a two-person college startup in a competitive, data-intensive vertical.

The episode's closing segment doubles as both practical advice and philosophical reflection. Yali's top recommendation is simple: find a co-founder who can support you when things get hard, because they will. Eyal adds a more analytical note — be scientifically honest with yourself about whether your idea has sufficient demand. Proving it to yourself first makes it exponentially easier to sell the vision to investors and future team members. Pat closes with a clean synthesis of the episode's thesis: most founders struggle with distribution while getting the product right, but Eyal and Yali's story flips that — they cracked distribution first and had to learn the hard way that product matters just as much. Both pillars are non-negotiable for building something that lasts. Pat closes with a final plug for Starter Story Build before signing off.