MIC

Mic — Hinge for Comedy Bookings Inspiration I am a comedian and producer based in New York. Through Jungle Cat Comedy and years of performing across the city, I lived the problem Mic solves firsthand. Comedians spend more time chasing gigs than performing them — cold emails, Instagram DMs, favors called in. Venues have the exact same problem in reverse. A bar in Astoria needs a storytelling comic for Friday nights. A cultural space in Harlem wants identity-based comedy. A supper club on the Upper West Side needs someone polished enough for a wine crowd. Neither side has any real infrastructure. No platform. No data. Just friction. That is what inspired Mic.

What It Does Mic is a two-sided matching platform for comedians and venues — built for both sides of the stage. Comedians build a profile with their style, draw size, minimum pay, availability, and clips. Venues do the same — capacity, pay structure, lighting and sound specs, previous performers, and what kind of comedy their crowd wants. Our AI matches them on what actually matters. Both sides swipe right using a mic button. When it is mutual, the match fires, an AI-generated icebreaker opens the conversation, and a booking is confirmed with a date that works for both parties — all inside the app.

How We Built It Mic was built entirely in React using Claude as both the development environment and the AI engine powering the product.

Matching engine — Claude Sonnet via the Anthropic API, prompted with full comedian and venue profile data to return a match score and explanation Search — Linkup API for real-time web search on comedian profiles so venues can verify talent instantly Messaging — A shared two-sided conversation model where both parties see the same thread from their own perspective Booking — An availability overlap algorithm that surfaces real upcoming calendar dates that work for both sides

jsconst overlap = comedian.availability.filter(d => venue.availability.includes(d)); The match score is a function of four signals: Match Score=f(style fit, drawcapacity, pay overlap, availability)\text{Match Score} = f\left(\text{style fit},\ \frac{\text{draw}}{\text{capacity}},\ \text{pay overlap},\ \text{availability}\right)Match Score=f(style fit, capacitydraw​, pay overlap, availability)

Challenges We Ran Into The two-sided inbox was the hardest problem. Early versions created duplicate conversations — one from each side. The fix was a single shared conversation object that both profiles read from, with perspective flipped based on which side is viewing. Image size limits hit hard. Embedding full-resolution photos as base64 pushed the artifact to 1.8 MB1.8\text{ MB} 1.8 MB, well over the render limit. Compressing images to 35% quality at 400px width brought the total payload down to 180 KB180\text{ KB} 180 KB — a 10×10\times 10× reduction.

The mutual match mechanic required tracking one-sided likes in a separate state array and only triggering the AI match call when the second party also swiped right — mirroring how real booking interest actually works. Prompt engineering for match explanations took several iterations. The final prompt instructs Claude to reference specific data points — draw versus capacity, pay versus minimum, style versus crowd — so the output reads like a real booker's recommendation rather than a generic AI response.

Accomplishments That We're Proud Of

Built a fully functional two-sided marketplace in a single hackathon session with no prior coding experience Integrated three AI systems — Anthropic for matching, Anthropic for icebreakers, and Linkup for live comedian search — into one seamless product Designed a mutual swipe mechanic that mirrors real booking dynamics: no venue is approached unless they have already expressed interest Built a date-picking system that surfaces real upcoming calendar dates based on overlapping availability from both parties The app works as a live demo — real matches, real messages, real booking confirmations

What We Learned Building Mic taught me that a two-sided marketplace is only as good as its matching logic. The UI is the easy part. The hard part is defining what a good match actually means — and translating that into a prompt that an AI can act on consistently. I also learned that AI is not just the product feature. It was the co-builder. Every component of Mic — the card layout, the messaging architecture, the booking logic — was built in conversation with Claude. For a comedian and producer with no engineering background, that changes everything about what is possible to build.

What's Next for Mic

Expand beyond New York — every city with a live comedy scene has this problem Real comedian profiles — let any emerging comic in NYC sign up and build their own card Venue accounts — let real bars, clubs, and cultural spaces list their rooms Revenue model — a small percentage fee on every confirmed booking, charged at the point of confirmation Analytics dashboard — show comedians their match rate, which venues viewed their profile, and which styles are most in demand in their borough Expand to other live performance verticals — musicians, spoken word artists, open mic poets — the matching infrastructure works for all of them

Built at the Anthropic Hackathon, April 2026.

By Arish Jamil — comedian, producer, and first-time AI builder. Annie Zheng - Engineer

Built With

• claude