Nepmind Labs
Voice-controlled AI chessGamingCompletedGamingAIVoice

Voicemate

Voice-controlled chess with an AI opponent—accessible play that keeps strategy and learning at the center.

Project snapshot
Client
Voicemate
Type
Startup
Region
North America
Engagement
MVP Build
Duration
5 months
Started
Q3 2025
MobileBackendAPIData
Objective

Ship a hands-free chess loop that feels natural for beginners, powerful for hobbyists, and welcoming for players who rely on voice — without sacrificing rules integrity or coachability.

Audience
Kids and families learning notation through speechPlayers with motor or vision differences who prefer voice inputCommuters and cooks who want a board without touching glassClubs and educators demonstrating inclusive play formats
Platform
MobileBackendAPIData
Problem

Touch chess on small screens punishes imprecision; traditional apps assume nimble fingers and quiet rooms. Many curious players never get past the friction of tiny pieces and abstract coordinates.

Solution

We paired a deterministic rules engine with low-latency speech intent: say the square, say the move, or ask the coach — the board updates instantly with clear narration, undo safety rails, and optional AI difficulty ladders.

Showcase

A closer look at Voicemate

Selected screens across the product — from marketing surfaces to in-app flows.

Mobile
Native iOS & Android flows
Welcome & voice primer
Voice-ready board
Mid-game play
AI & coach settings
Learn & practice
Profile & accessibility
Features

What we built

The core capabilities shipped as part of the engagement.

Voice-first move input with confirmation and disambiguation
Adjustable AI opponent with explainable hints and blunder alerts
Coach mode: legal moves, threats, and simple plans read aloud
Accessibility presets: larger pieces, high contrast, haptics, pacing
Lesson tracks that link voice commands to opening and tactic ideas
Local and online game history with PGN export and replay narration
Architecture

How it's built

The stack and architectural layers powering the product.

Core stack
FlutterNode.jsPostgreSQLAWS

Backend

Node.jsExpress

Mobile

FlutterOn-device speech (primary path)WebSocket client

Data

PostgreSQLRedis

Infrastructure

AWS ECSS3CloudWatchGitHub Actions

Integrations

Platform speech recognition APIsStockfish-class engine serviceOptional cloud ASR fallback
Compliance & security
COPPA-aware defaults for under-13 modesVoice audio minimization and configurable retentionEncryption for account and game history
Delivery

Timeline & team

How the engagement was structured across 21 weeks of work.

Phase 1
2w
Discovery
Phase 2
14w
Build
Phase 3
5w
QA
Team on this engagement
Product ManagerUI/UX DesignerFlutter EngineerBackend EngineerML-adjacent engineer (engine + ASR integration)QA Engineer
Impact

Results that mattered

Measurable outcomes tracked against the KPIs that drove the build.

Voice move success rate
91% accepted on first utterance
Session length
+34% vs. touch-only beta
Accessibility opt-in usage
28% of weekly actives
Outcomes

What changed for the business

Launched cross-platform chess with shared rules engine contract
Reduced ambiguous move handling with structured clarification UI
Established inclusive UX patterns adopted across other Nepmind game concepts
“Nepmind made voice chess feel obvious — not gimmicky. Our play-testers with limited dexterity finally have a board that listens to them.”
Dr. Leah Morrison
Founder, Voicemate

Have a project like Voicemate?

Tell us what you're building — we'll help you ship it with the same rigor.

Start a projectBrowse more work