GAMEPLAN

You can't practice against Sentinels without playing Sentinels.

Every coach knows this. You study VODs, you build reads, you theorize counter-strats. But testing them? That takes 10 players and blocked scrim time.

GRID gives us VCT telemetry — where every pro stood, every millisecond. We turned that into AI agents that execute like the players they're modeled on.

Gameplan lets you run the round before match day.

This is Moneyball for VALORANT. One player out of position. One late rotate. One missed trade. Gameplan connects micro-level mistakes to macro-level outcomes — so coaches can see exactly where rounds fall apart and what to fix.

What Gameplan Does

A comprehensive assistant coach — not a report generator, but a simulation engine that lets you test strategies before match day.

Feature	What It Does	Why It Matters
Run Simulation	Run rounds from scratch on any map	Test your own setups without loading VCT data
VCT Match Replay	Replay any pro round with full position data	Study exactly what happened, frame by frame
What-If Simulation	Re-run rounds with different strategies	Compare outcomes without needing 10 players
Strategy Planner	Design executes anchored to pro behavior	Plans grounded in real VCT patterns
AI Decision System	Agents rotate, trade, and clutch like pros	Calibrated on 592K+ VCT position samples
AI Coaching Chat	ASI1:mini-powered tactical analysis	Ask "why did this round fail?" — get insights

Who Benefits

Stakeholder	Gameplan Value
Head Coaches	Simulate opponent tendencies, test counter-strats before match day
Analysts	Jump straight to tactical scenarios instead of drowning in VOD
Players	Visualize setups before scrims, understand why calls are made

The Foundation

What We Processed	Volume
VCT Position Samples	592,893
Kill Events	12,029
Player Profiles	85
Maps	11

Core Systems:

Simulation Engine (Python/NumPy) — core round simulation logic
A* Pathfinding with visibility — agents avoid sightlines, use cover
Combat Model — damage falloff, accuracy curves, trade timing
Economy Engine — eco/force/full-buy decisions
Ability System — utility timing and placement from pro patterns

Stack: Next.js 16 · React 19 · FastAPI · PostgreSQL · Redis · Pixi.js · GSAP · Zustand · TypeScript · TailwindCSS 4

Built With

GRID.gg (VCT telemetry) · ASI1:mini (AI coaching) · JetBrains + Junie (AI development) · Vercel (frontend) · Google Cloud (backend)

Links

Resource	URL
Live Demo	https://c9-gameplan-frontend.vercel.app
Backend API	https://c9-gameplan-backend-902522310828.us-central1.run.app
Frontend Repo	https://github.com/joshghal/c9-gameplan-frontend
Backend Repo	https://github.com/joshghal/c9-gameplan-backend