← Back to projects Case study · 48‑hour build

Datacoach

An AI coaching companion that turns a smartphone and a Snapchat lens into a personal sports tutor — built in a weekend, Top 10 at Founders Inc, sponsored by the Golden State Warriors.

Top 10 · Founders Inc Sponsored by GS Warriors 2023

The problem

Most kids don't have access to professional sports coaching. Pro tools — pose tracking, game film analysis, biomechanics breakdowns — live behind tens of thousands of dollars in equipment and a coach who knows how to read it. Meanwhile, parents at home have a smartphone, a driveway, and a kid who wants to get better.

The gap between "I have a phone in my hand" and "I have actionable feedback" is enormous. We wanted to close it in a weekend.

The approach (the 48‑hour build)

The whole thing was built in plain English over a Founders Inc weekend — no engineering team, no traditional code. The trick was stitching together two consumer technologies most people overlook:

  • A Snapchat AR lens that uses the platform's built‑in body tracking to capture pose, form, and movement from a normal phone camera — no special hardware required.
  • A conversational bot on top of an LLM that takes the AR lens output and translates it into coach‑style feedback: drills to try, what your shoulder is doing wrong, what to focus on next.
  • A glue layer that pipes the two together so a parent can record a single rep and get a paragraph of personalized advice back in seconds.

The bet: AR + LLM is enough to get 80% of the value of a real coach for 0% of the cost. The judges agreed.

The result

  • Top 10 finalist at the Founders Inc hackathon out of a stacked field.
  • Sponsored by the Golden State Warriors as one of the standout sports‑tech entries of the weekend.
  • Published on Devpost with a working demo video that still gets traffic from search.
  • Validated the broader vibe‑coding thesis: a non‑engineer with the right prompts can ship a finalist‑grade prototype in a weekend.

What I'd do next

  • Ship a real iOS / Android app with on‑device pose tracking so it works without leaving Snapchat.
  • Train the conversational layer on actual coaching transcripts per sport (basketball, soccer, tennis) so the feedback feels less generic.
  • Add a parent‑facing weekly progress report — the kind of artifact a school or AAU team would actually pay for.
  • Open it up to youth sports leagues as a low‑cost tier so coaches can drop in personalized notes per player without burning their evenings.

Want to talk about building something like this?

I run workshops and one‑on‑one sessions on shipping AI products in a weekend.

Book 15 min