AI GamingGPT & Claude

VoiceCue — AI-Powered Voice-Triggered In-Game Advertising

Our primary users are independent game studios and multiplayer sandbox game operators that need sustainable monetization without disrupting gameplay. Our secondary users are advertisers seeking relevant engagement and gamers who want immersive, uninterrupted experiences. Today, in-game advertising is largely disconnected from what players actually want. Static billboards, generic promotions, and intrusive pop-ups target exposure rather than real-time player intent. A player might say, “I'm starving; I could really use a pizza,” while the game has no mechanism to recognize that moment as a relevant advertising opportunity. Meanwhile, advertisers spend money reaching players who may have no interest in their offers. For smaller game studios, building AI-powered advertising infrastructure is technically complex and costly. They lack an accessible way to connect player-expressed needs with contextual sponsorships, measure engagement, and generate additional revenue. The result is a three-sided problem: studios miss monetization opportunities, advertisers struggle to reach genuinely interested audiences, and players encounter irrelevant advertising that breaks immersion. Any solution must also respect player consent and privacy, especially when voice data is involved.

GitHub

Preliminary score

50.5/ 100Earlyaverage of 2 judges

Value for the user15.0 / 25
A specific problem and a clear outcome from solving it.
Prototype and use of AI15.5 / 30
A working core scenario, and what AI actually contributes to the result.
Quality testing6.5 / 20
Tests, examples of failures, and comparison with the current approach.
Feasibility7.5 / 15
Data requirements, running costs and a clear next step.
Originality6.0 / 10
A meaningful difference from common solutions.

What the judges wrote

GPT-6 AstraPreliminary55/100
Strengths
Your documented voice-to-intent-to-billboard workflow addresses a specific monetization problem for small game studios without relying on interruptive pop-ups. Your campaign portal, auction logic, and proximity-based offer interaction make the concept more concrete than a standalone recommendation tool. Your explicit disclosure of rule-based intent detection and simulated commercial metrics usefully distinguishes prototype behavior from proven advertising impact.
To improve
Your prototype evidence would be stronger with a reproducible speech-to-billboard run showing actual event outputs for both simulator and microphone paths. Your testing should report labeled phrases, false-trigger rates, negation and ambiguity failures, and player-experience comparisons against static ads. Your on-device privacy guarantee needs validation because browser Web Speech recognition may use remote services; also test consent withdrawal and session expiry. Your feasibility plan should quantify serving costs and define a small studio pilot with measurable opt-in, engagement, and player-disruption outcomes.
Claude Opus 5.5Preliminary46/100
Strengths
Your prototype covers a lot of ground: a walkable Three.js world with billboard slots, an advertiser portal, a studio console with brand-safety filtering, and a Vickrey second-price auction with a defined 70/30 revenue split. Your privacy design is a thoughtful touch for a voice-based ad product: speech is reduced on the client to coarse intent tags, with opt-in consent and a 10-minute session TTL. You were honest about the limits, stating plainly that intent detection is keyword-based, metrics are simulated and no LLM or ML model is used.
To improve
The AI contribution is thin. Apart from the browser's speech-to-text, intent detection is regex and keywords, so a small embedding or LLM classifier that handles ambiguous phrases would make AI essential rather than incidental. Your testing section describes deployment fixes, not quality checks. Build a set of 50–100 spoken phrases, report intent accuracy, false triggers and failure cases, and compare click-through against a static billboard baseline. Add running cost estimates and one concrete next step, such as a pilot with a named indie studio, and give judges a working demo link.

The team

Fidan Baghirova
data
Emil AkbarovLead
developer
Fatima Alibabayeva
data

Disclosure

Required by the rules. Judges verify it against the repository.

Models: Browser Web Speech API (speech-to-text); custom JavaScript keyword-based intent classifier; Python regex-based intent extraction and heuristic content moderation. No external LLM or pretrained ML model.
Data: Synthetic demo data and predefined test scenarios. Seeded advertiser campaigns, example product categories, in-game advertising slots, sample speech phrases, simulated impressions, clicks, revenue metrics and moderation cases. No external customer dataset or proprietary training data.
Components: Three.js (WebGL 3D world), Vite, vanilla JavaScript, HTML/CSS, Node.js, Express, Python, FastAPI, Uvicorn, Pydantic, UUID, CORS, Docker, Docker Compose and Render deployment configuration. Custom auction engine, privacy gateway and in-memory session store.
Built during the event: Confirmed

Quality testing

We tested the end-to-end workflow, including voice-triggered intent detection, campaign matching, dynamic in-game advertisements, and user interactions. We also checked content filtering and advertiser management. Our main challenge was deployment: dependency conflicts and Express routing errors initially prevented the application from running correctly on Render. We resolved these issues by adjusting dependencies and server configuration. We identified limitations in browser-based speech recognition and keyword-based intent detection, especially with ambiguous phrases. Compared to traditional static in-game ads, VoiceCue demonstrates real-time, context-triggered advertising within gameplay. Our results are prototype-level, with simulated commercial metrics rather than real conversion data.

Worth 20 of the 100 points.

Running it

Access: Open the provided demo link in Chrome or Edge on a desktop computer. No registration, credentials, or API keys are required. Suggested Demo Walkthrough (2–3 minutes): 1. Enter the 3D Game World: Click "Enter Game World" to explore our interactive Minecraft-inspired environment. 2. Trigger Contextual Advertising: Enable Smart Suggestions and select "Want pizza" from the built-in speech simulator, or use the microphone with browser permission. Watch the system detect the player's intent and dynamically update an in-game advertising billboard. 3. Interact with the Advertisement: Approach the sponsored billboard and press E to view the promotional offer and in-game reward. 4. Explore the B2B Advertiser Portal: Inspect active campaigns, configure targeting categories, and explore the campaign creation workflow. 5. Explore the Studio Console: Review advertising inventory, revenue-sharing logic, and brand-safety moderation controls. 6. Inspect Architecture & Telemetry: Follow the event pipeline from player speech to intent classification, advertising decisions, and in-game rendering. Known Limitations: The prototype uses rule-based intent classification and simulated advertiser data. Live microphone recognition depends on browser support. No real purchases or commercial advertising transactions are processed. Tip: The built-in speech simulator provides the most reliable demonstration without requiring microphone access.

Project ownership

The team keeps the rights to this project. Judges were given access for verification only.