Skip to main content
T
TYSONLABS
Back to software directory
AI & Agents·Open SourceAI Chess Coach

Tyson Chess · Interactive AI Chess Coach

AI chess mentor that spots your blunders and teaches you how to win in plain English.

Stack:ReactTypeScriptAI Chess EngineWeb WorkersPWATailwind CSS
System Topology & Operational Flow
Tyson Chess · Interactive AI Chess Coach
AI & Agents
01 / UI Surface
Zero-latency React board
02 / WASM SIMD
In-browser Stockfish 18
03 / Analysis Engine
24-ply tactical evaluation
04 / AI Coach
Plain-English move guidance
Architecture: Sovereign Codebase · Sub-10ms Latency
Ownership: 100% Client Sovereign IP
Verified Status: October 2026Zero employer conflicts · Independent IP
BUSINESS OPERATIONAL CONTEXT

Why I built this & what it solved

Vertical: Chess Academies, Intermediate Learners & EdTech
The Origin & Why I Built It

I used to play on the Chess.com app every day, and for solving more than 3 puzzles a day you needed to pay ₹99 per week. Paying a recurring subscription just to practice chess tactics felt absurd. So I thought: why not build it myself?

The Core Problem Solved

An independent in-browser chess academy that runs an intelligent engine directly on your device with zero server latency. It provides unlimited tactical puzzles, replaces confusing computer numbers with plain-English coaching ("Knight to d5 forks Queen and Bishop"), and offers 6 calibrated human-style sparring bots — 100% free forever.

Measurable Business Impact:
Unlimited daily tactical puzzles without a ₹99/week subscription paywall
$0/month server cost running 100% locally in-browser
Plain-English tactical coaching replacing cryptic numeric evaluations

Overview & Problem Space

I used to play on the Chess.com app every day, but for solving more than 3 puzzles a day or getting basic move reviews you had to pay ₹99 per week (₹5,000+ per year). Paying a recurring weekly subscription just to practice chess tactics felt ridiculous. I thought: why not build it myself? Tyson Chess was engineered to provide unlimited tactical puzzles, plain-English blunder coaching, and human-style sparring bots with zero paywalls.

The Operational Friction:

Commercial chess apps lock daily tactical training behind aggressive paywalls (₹99/week) while spitting out raw, robotic engine evaluations (+1.84) that fail to explain why a move actually lost or won.

Target Operator:

Club players, adult improvers, and intermediate learners seeking unlimited puzzle practice and intuitive blunder coaching without paying weekly subscriptions.

The Solution & Contribution

A zero-backend web platform running intelligent chess AI in dedicated web workers. It features a blunder safety net that halts catastrophic moves during casual play and prompts tactical reflection.

Direct Responsibility & Engineering Role:

Creator & Engine Engineer — Compiled and tuned the WebAssembly engine integration, built the custom bot heuristic softmax weighting, and created the spaced repetition blunder scheduler.

Technical Architecture

Frontend Layer
React, TypeScript, react-chessboard, Tailwind CSS, Lucide icons
Backend & Routing
Zero-backend client architecture with localStorage schema versioning
AI & Heuristics
Stockfish AI Engine with custom evaluation and coaching wrappers

Key Features & Capabilities

Plain English Tactical Coach

Coaching

Replaces raw engine numbers with explicit tactical callouts: "Knight is pinned to the King", "Rook hangs without defenders".

Calibrated Softmax Bots

Engine

Bots simulate human mistake profiles rather than robotic depth cutoffs, making tactical slip-ups characteristic of their Elo band.

Personal Mistake Vault

Spaced Repetition

Automatically captures every blunder made during sparring and converts it into a scheduled puzzle review.

Repertoire & Trap Trainer

Curriculum

11 annotated master opening repertoires and 12 classic traps with interactive play-and-defend tests.

Engineering Decisions & Insights

#01

Used high-speed web worker multithreading with graceful single-thread fallback for universal device compatibility.

#02

Implemented hash-based routing (`/#/play`) to allow static PWA installation on iOS and Android with zero server routing requirements.

L1

Balancing bot difficulty required empirical round-robin tournament simulation scripts to verify that rating ordering remained consistent across thousands of games.

Operational Benchmarks

Stockfish AI
Coaching Engine
Runs in-browser
230+ Puzzles
Drill Library
Mined from bot blunders
$0/mo
Backend Cost
100% client execution

Related Systems

View all 16 builds →
#01/SaaS & Industry OS
Live in Production

DinePulse · Restaurant & Hospitality OS

High-speed 3-click restaurant billing till with table orders, live kitchen screens, and instant UPI settlement.

Tactile billing till built for busy restaurant rushes. Table orders, kitchen order tickets (KOT), Indian FSSAI veg/non-veg tags, and automated WhatsApp receipts.

3 clicks from order selection to receipt generation · <300ms kitchen order dispatch latency
Next.js 15TypeScriptTailwind CSSRazorpay UPI+2
#06/AI & Agents
Open Source

Chief of Staff AI · Autonomous Multi-Agent Framework

Autonomous AI engineering team for research, coding, testing, and operations that works with zero hallucination.

Multi-agent orchestration system pairing reusable workflow skills with specialized personas (orchestrator, engineer, reviewer) for autonomous 24/7 execution.

60–90% reduction in agent token costs via compression · Zero unverified code merges with automated quality gates
PythonNode.jsAutonomous AgentsGit Hooks+1
#11/AI & Agents
Prototype

ReviewAssist AI · Google Reviews Reputation Engine

AI-assisted reputation management converting positive feedback into Google reviews.

Frictionless customer review collection SaaS using LLMs to draft personalized reviews while complying strictly with FTC & Google policies.

+42% increase in authentic 5-star Google Maps reviews · 100% compliance with FTC and Google anti-incentivization rules
Next.jsNestJSOpenAI API (GPT-4o-mini)PostgreSQL+2