Development proposal · AI product
An app that tells every coach which play to run, which drill to work, and which video to watch — and even helps create brand-new plays to sell. Coach D's playbook, guided by AI.
Prepared for Complete Coach D · Scope: MVP — Guided recommender
The product in one screen
No scrolling through hundreds of plays: the AI reads the game situation, picks from Coach D's library, and explains why — with the diagram, the drill and the video to watch it.
The blitz opens the middle. The fake to B freezes the contain; C attacks the post with leverage. High-percentage call on 3rd & long.
How it works
Format (5v5 / 6v6 / 7v7), level and age group. Once — the app remembers your context.
Down and distance, field position, offense or defense, and the opponent's tendency.
Plays and formations with the diagram, the why, the drill, and the video to watch it.
Module 2 · Play generator
Describe a play in plain words and the AI drafts it — formation, routes, name and type — in the very format the app already draws and animates. You refine it in the existing editor, approve it, and it becomes a real, sellable play. You always have the final say.
“6v6 goal-line, fake handoff, two crossing routes.”
A full first draft: formation, routes, name and play type.
Tweak it in the editor you already have. Nothing ships without your OK.
It becomes a real play — ready to use and to sell.
Three revenue engines: coach subscriptions for the recommender, a premium tier where coaches pay to generate their own plays, and every play you generate becomes fresh inventory for your store — your Shopify is already connected. The AI turns into a content factory: new playbooks in hours, not days.
The head start
We're not starting from scratch. We audited the current platform (PlayBuilder) and found that most pieces already exist — that lowers the cost and shortens the timeline.
Bottom line: the project is mostly about connecting and shaping what already exists — the AI, the plays, the videos — inside a modern app. Less risk, lower cost, faster to production.
Architecture
Exactly as envisioned: a new, decoupled front end on top of the current system as its data backend. No rewriting what already works.
iOS + Android + Web from one shared codebase.
Orchestrates the AI engine and picks the relevant plays.
Laravel + MySQL: plays, playbooks, users, videos.
Platform
Installable as an app (PWA) on any device.
Native feel and push notifications.
Same app, same code, both stores.
With Expo (React Native) all three ship from one shared codebase · option: launch web-first and package to the stores later.
Timeline
Six phases, from the line of scrimmage to the end zone. Phases 2 and 3 overlap; content runs in parallel.
Investment
Both modules — the recommender and the play generator — for a single fixed price, plus a low monthly for operation. Affordable up front, and built to pay for itself.
Recommends the right play, drill and video from Coach D's library — for every game situation.
Create brand-new plays to sell — plus the premium tier where coaches generate their own.
One-time · 50% upfront / 50% on delivery · delivered in ~8–10 weeks.
+ $100 USD / mo for operation & support (AI usage, hosting, minor tweaks; cancel anytime).
App stores (paid by the client): Apple $99/yr, Google Play $25 one-time — the current VPS already runs the backend, so no new hosting. Return: three recurring revenue streams — coach subscriptions, the premium play-generator, and selling the playbooks you generate — and since the videos already exist on YouTube, there's no content-production cost.
Transparency
“Flag Football with Coach D” on YouTube. Integrated and cataloged — no production.
The ~540k plays already carry type and format; a direct basis for the recommendations.
Quality depends on documenting how Coach D decides. His input in Phase 1 is key.
The price covers both modules — recommender and play generator. Generated plays always pass through your review before publishing. Conversational chat and opponent analysis are later phases.
Next step
The foundation is clear and the biggest unknown — the videos — is already solved. Just three decisions from the client: