Building a GenAI Product From Scratch for a Traditional Sports-Betting Industry — and the Pivot That Followed - Product Builder Podcast episode with Eyal David
Updated:

Building a GenAI Product From Scratch for a Traditional Sports-Betting Industry — and the Pivot That Followed

Host: · Guest: Itzik Adziashvili · About the host

In real-time data, you can't maximize coverage, speed, and accuracy simultaneously — balancing them is what makes the job hard. According to Itzik Adziashvili, GM and veteran product manager at LSports, a bootstrapped company feeding data for 3 million matches a year to betting sites and media, that balance — coverage, sub-second latency, high accuracy — is the competitive edge.

In this episode

  • The sports-betting industry is traditional and slow to adopt technology (an integration can take half a year) — but that's exactly where the opportunity lies: bringing innovation and getting ahead of the veteran players who focus on the commercial side.
  • LSports' whole game is a constant balance between three axes — coverage, latency (sub-second), and accuracy. Favoring one axis is easy; balancing all three is what forces the hard work.
  • The two most important skills for a product manager in the data-pipeline world: a deep connection to the technology with attention to detail, and the ability to 'manage focus like an accordion' — drilling down to the single event and zooming out to the whole picture.
  • A culture of direct communication is built on trust: cut to the chase, short 1-2-3-person meetings, make a reasoned decision and move forward instead of stalling — and sometimes just lowering your ego to reach the right decision faster.
  • A GenAI product isn't built like a normal MVP: because the B2B chat is shown to super-professional clients at the conference, they went for a mature, sellable product rather than a demo — and they measured accuracy through end-user conversion, not just model metrics.
  • After the conference hype came a pivot: client adoption barriers (fear of injecting a chat into their site, guardrails, languages and slang) drove a move from a widget toward API-based/social — and also opened a new business line they hadn't thought of.

Episode Description

Itzik Adziashvili, a GM and veteran product manager at LSports, tells the story of how a bootstrapped company of nearly 400 people builds the data infrastructure that feeds sports-betting sites and media worldwide — 3 million matches a year across 60 sports, at sub-second latency and high accuracy. The conversation moves from the built-in tension between coverage, speed, and accuracy, through a direct, bullshit-free work culture, all the way to the full story of building a GenAI product from scratch for the annual conference — including the pivot that followed.

Read the full transcript
Full transcript with speakers and timestamps

Selected quotes

“If I tilted toward one of my axes, my life would be much easier — to give as much as possible at the expense of accuracy, or to give it as fast as possible; when I have to balance between them, that's exactly the world that forces us to work much harder.”
— Itzik Adziashvili
“Let's get to the service, to what's needed — tell me how I can help you with fewer stories.”
— Itzik Adziashvili
“Normally, when a product is at fifty percent accuracy and it's version one, I as a product manager would go for it — I'd launch and start rolling it out.”
— Itzik Adziashvili

Questions & answers

What does LSports actually do?

It provides real-time sports data to sports-betting sites and media (B2B). It collects huge volumes of data, normalizes it, and delivers it through an automatic, accurate pipeline at very low latency, so the client can focus on optimizing their own users.

What is the company measured on by its clients?

Two main axes: coverage — how many matches and events can be supplied at any given moment; and latency — updates fast down to the second, per the SLA. All while maintaining high accuracy, where the differences are measured in per-mille.

Why did a company in a traditional industry bet on a GenAI product?

To ride the trend and get ahead of the market: on one hand a traditional industry, on the other the cutting edge of technology. They wanted to test whether it was just buzz or real value, and used the window of the annual conference to bring an innovation.

How did they measure the AI product's quality?

They started with the accuracy of scientific tagging and parameters like precision/recall against manual tests, but realized it's hard to distill real AI-evals. So they moved to measuring value through end-user conversion — on the assumption that if the user stayed and completed an action, the client got value.

What happened to the product after the conference?

It created a lot of hype and excitement, but client adoption barriers emerged — fear of changing sites at a high pace, guardrail issues, users' languages and slang. As a result they pivoted from a widget toward API-based/social, which is easier to consume.

Product Management Sports Betting Real-Time Data Latency GenAI LSports Bootstrap Product