← TrendWatcher
GitHub Trending
5/10

DrillLab

A phone app that lets amateur soccer, basketball, and hockey players practice game scenarios in a continuously generated video world — try a free kick against an AI-generated keeper, experiment with tactical choices, save the clip.

Target user

youth and amateur sports players and their coaches

Features
  • scenario library (corner kicks, breakaways, defensive switches) with live video responses to your choices
  • shareable short clips of your 'plays' for team group chats
  • weekly training plans suggested from the scenarios you struggled with
  • coach mode that lets a coach stage a drill and review each player's clip submissions
Why now

MIRA demonstrates that multiplayer world models can run at 20 FPS on a single GPU, which puts real-time consumer sports simulation within reach for the first time.

Signals · overall 5/10
Demand
6/10

HomeCourt is a real consumer basketball/soccer training app used by amateurs and youth programs (Soccer Stars partnership), and AppsHunter lists 188+ sports-simulation apps, showing live demand — but no evidence yet of demand for *generated-video-world* practice specifically.HomeCourtHomecourt - Soccer Stars188+ Best Sports Simulation Apps & Games for iPhone (2026)

Whitespace
7/10

Adjacent competitors (HomeCourt, MOTI 3D Play Sim) use computer vision on real video or coach-facing 3D, not generative video worlds for amateur practice; MIRA/Matrix-Game are research demos — niche is open but not uncontested.MOTI Sports Simulator Software - 3D Play Simulation ToolMatrix-Game 3.0: Real-Time and Interactive World Model

Monetization
5/10

Sports training apps monetize via team/club subscriptions (Hudl, TeamSnap, GameOn, Futures App target organizations); willingness of individual amateur youth players or parents to pay for a generative-world practice app is unproven.Hudl PricingTeamSnap Pricing for Clubs, Leagues, Coaches, & TeamsPricing | GameOn! Sports App

Longevity
6/10

AI sports training is a durable trend, but the specific claim that MIRA 'puts real-time consumer sports simulation within reach' overstates maturity — model is 5B params at 288p/20fps on a single GPU, far from phone-ready.MIRA - Blog postGitHub - mira-wm/mira

Feasibility
3/10

MIRA needs a single datacenter GPU and is a 5B-parameter model at 288p; running a multiplayer latent diffusion world model for soccer/basketball/hockey on a phone with usable latency is a major unsolved systems problem, not a near-term shipping product.GitHub - mira-wm/miraMIRA - Blog post

Code for MIRA: Multiplayer Interactive World Models with Representation Autoencoders · ★ 110GitHub Trending · 2026-07-07 (18d ago)