TISC Is Giving Coaches the Power to Simulate the Next Play

Written by Sarah Willson, Arpan Bhattacharya and Mahbod Sabbaghi

A college basketball team just earned its first tournament invite in 23 years. Behind that turnaround was a coaching staff that spent the season working closely with TISC, a Toronto and SF-based startup building world models for coordinated human movement. 

Panache Ventures participated in TISC’s US$2.1M pre-seed round alongside OVO Fund and N49P. At the college and pro level, sports games are often filmed from multiple angles, with every movement tracked. Most sports-tech platforms today use that data to help coaches understand what happened and plan for the next game. Yet many of the highest-leverage decisions are still made on instinct, in real time. We invested in TISC because we believe the next edge in sports will come from helping coaches make better decisions in real time using live human movement data. Elite athletes represent human movement at its most refined, which makes sport the best place for AI to learn how people move and interact. That same capability applies anywhere people or machines need to coordinate movement, from robotics to defense to search and rescue. We believe TISC has the opportunity to become the leading intelligence layer for human movement and coordination, starting with sports. 

Coaches spend countless hours breaking down film, analyzing opponents, drawing up plays and running scenarios in practice. Despite decades of expertise and months of preparation, coaches cannot anticipate every possible scenario, especially in real time game play with the clock running. TISC turns game film into a multi-agent world model, helping coaches anticipate how a game will unfold and choose the best response in real time. The model is trained on game film and player tracking data from more than 40 college and pro teams. TISC learns the dynamics of play, how each player and the ball move, and how every player reacts to the others around them. It then generates realistic video of those predictions, governed by real physics and tactics. A coach can pause any moment, redraw a route or sub in a different player, and watch the next eight seconds play out. TISC’s generated plays look close enough to real game footage that coaches can watch them like film, with players colliding, blocking and tackling the way they would in a real game. 

For example, imagine a football coach wants to simulate how a play might unfold in real time. Drawing on historical game data and real-time insights into how each player moves and behaves, TISC can generate more than 1,000 potential scenarios and outcomes in seconds. It can then turn any of those scenarios into realistic video that looks just like a continuation of the original game footage.

College sports alone represents a significant opportunity for TISC, as programs pour more into coaching and competitive preparation than ever before. Division I programs spent $20.8 billion on athletics in the 2023-24 academic year, Sports analytics specifically is estimated at roughly $2.3 billion globally in 2025 and projected to reach $4.75 billion by 2030. Today’s coaching tools are largely retrospective, built to review film, track performance and scout opponents after the fact, while the decisions that shape a game in real time still rely almost entirely on experience and instinct. TISC is building for that gap, giving coaches a predictive view of what is likely to happen next and how best to respond.

Building for this gap takes a mix of frontier AI research and deep sports expertise, which is exactly the team TISC’s co-founders have assembled. Arpan Bhattacharya and Mahbod (Moe) Sabbaghi have been hacking on problems together since meeting at Wish in 2021. Arpan went on to work across infrastructure, product and machine learning at Meta and Uber, while Moe studied computer science at Waterloo. In 2024, Arpan left Uber and Moe dropped out of Waterloo to start TISC, and they have since attracted researchers in multi-agent coordination and world models from Mila and Meta, as well as engineers who built search at Shopify and chips at AMD. They have also brought on sports insiders, including an NBA Hall of Famer and Finals MVP, a former co-GM of the Dallas Mavericks and a two-time NBA champion coach. 

TISC’s model is already 2.5x more accurate than Gemini at identifying basketball play types and ranks first against published models at predicting where every player and the ball will go. Sport may be the richest training ground for AI that understands human movement, with more film, tracking data and expert labels than almost any other domain. Long term, TISC’s technology has the potential to disrupt not only sports, but any setting involving the prediction of human movement.  

With TISC’s model outperforming industry benchmarks and teams seeing improvements from its real-time insights, the team is ready to scale. This next phase is about bringing more programs onto the platform, adding research and engineering talent, and securing the computer to train larger models across more sports and environments. If you’re an investor, coach, or operator who wants to help TISC get there, we would love to make an intro.

About TISC

TISC has built the most accurate world model for sports, developed alongside over 40 college and pro teams. Their platform allows coaches to see how a game is likely to unfold in real time and respond with confidence. Sport is where TISC is starting, but the applications of this technology extend far beyond athletics. Learn more at https://tisc.co/