Heisenroy

I built Outfix, a virtual wardrobe assistant that helps you "Fix your Outfit' by just talking to it. Like a stylist. I actually have been building this product out since week 2 for cloudflare and wanted to submit it for TurboPuffer (where i would store the clothing data), but then it took longer and eventually I decided to grind this out with Cursor, finalizing just today. Was finally able to use my domain that I bought some time ago :).

Submitted 14 May 2026

What if books on Goodreads can be summarized by an AI agent?

Duke — heads-up no-limit Hold'em against an AI you can actually talk to. A weathered Vegas pro plays real poker (deterministic strategy + opponent profiling) while a separate ElevenLabs Conversational AI agent voices him in character — free to bluff, lie, and trash-talk you in real time. Built with ElevenLabs Conversational AI and Sound Generation.

Reading Companion is a hands-free reading assistant that lets you look up words, set timers, and play ambient music — all by voice, without breaking your reading flow. When you encounter an unfamiliar word, just say it out loud and ElevenLabs Conversational AI speaks the definition back instantly. The same voice agent handles commands like "set a timer for 30 minutes" or "play lofi hip hop" using ElevenLabs client tools wired to real browser actions. The entire app was built with Kiro's spec-driven development workflow — requirements, design, and implementation tasks were all structured as Kiro specs, with Kiro executing each task sequentially from backend to browser client, using the ElevenLabs Kiro Power for accurate API guidance throughout.

AI-powered audio engine using turbopuffer vector search to semantically map natural language to rich audio landscapes. Inputs are embedded into a high-dimensional mood space, matched via cosine similarity, then synthesized through ElevenLabs' generative audio pipeline into immersive, context-aware soundscapes in real-time.

Ever wanted to hear your favorite artist's voice on a song you wish your artist was on? Or you just hear a voice so good that you want to see what it would sound like on your favorite song? Using ElevenLabs and Replit I built an application that clones your artist's voice on youtube/tiktok and then replaces it with the artist's voice on the song you want to sample. All done through Replit.

Ember is a therapeutic AI campfire companion built to address the accessibility gap in mental health support. Therapy is expensive, waitlists are long, and sometimes people just need someone to listen at 2am. Ember fills that gap — you click on an interactive 3D campfire built with Three.js, and a warm, supportive AI companion greets you by voice and has a natural spoken conversation with you while the fire reacts in real time, dimming when Ember speaks and flaring when it listens. The campfire metaphor is intentional: fire is universally calming and lowers the barrier to opening up compared to a clinical chat interface. ElevenLabs powers Ember's voice using the TTS API with the Brittney voice and the eleven_turbo_v2_5 model, proxied through a Cloudflare Worker endpoint that keeps the API key server-side. Voice is essential to the experience — reading text would break the campfire illusion, so Ember always speaks aloud. On the Cloudflare side, a Worker handles all routing including the ElevenLabs proxy and Agent SDK WebSocket connections, while a Durable Object gives each user their own persistent CampfireAgent instance that maintains full conversation history across sessions so Ember remembers returning users and greets them accordingly. Workers AI runs Llama 3.3 70B with a voice-optimized system prompt that keeps Ember warm, concise, and natural-sounding, all on Cloudflare's infrastructure with no external LLM API. Ember isn't a replacement for professional help — it includes crisis resource referrals like the 988 Lifeline — but it's a calming, judgment-free companion that's always there when you need it.

ChessAgent is a Chrome extension that gives chess.com players a real-time AI scouting report on their opponent before the game even starts. In blitz and bullet chess, you simply don't have time to open a new tab, look up your opponent's profile, dig through their stats, and figure out what they're good or bad at. ChessAgent solves this by automatically detecting your opponent the moment you're paired, scraping their full profile, recent games, openings, and loss pattern, and then using ElevenLabs to speak the most critical insight directly to you in a natural voice before you even make your first move. You never have to look away from the board. After the game ends, ChessAgent shifts into coaching mode: it analyzes the PGN of the game you just played, generates a short debrief about the moment that decided the outcome, and then uses Firecrawl to search the web and surface the most relevant video lessons for what you need to study. Most importantly, ChessAgent completely abides by fair play policy: it only uses publicly available information from chess.com's open API and public game archive.