Velvet Rope by Robot Rock

The analogy is a Bouncer, DJ, and an Event Organizer for an event at a venue. The tool allows you to easily sift through applicants for a potential event. We believe that the vetting process should still be done by the human hosting the event, for now (until you make a decent AI clone of yourself).

We allow you to interact with AI clones of your potential event attendees. We then generate and play “pleas” or “pitches” of your event applicants (simulated to represent them as accurately as possible given the given data - - currently with just voice pitches but maybe AI avatars in the future). You watch these pitches while we use AI-emotion detection tools to read your visceral expressions and microreactions to the applicant pitches for attendance - - this is you being the “bouncer” who you (or your emotions) choose to admit or not. In the future we could add other biomarkers like pulse and pupil dilation to determine your visceral reaction to these applicants.

After you set venue attendance parameters, we allow entry to the top emotional reactions.

Then, we let you craft the details and dynamics of the venue as the “DJ” who decides the dynamics and rules of the venue and event.

Then we create AI clones of your attendees and place them in the “venue” and simulate attraction and repulsive forces between members until they form groups and begin working together.

After they form groups, they begin to discuss and produce an “output” based on the venue, DJ rules, and dynamics.

At the end we cross reference the outputs with the stated user goal of the overall event.

This allows you a new AI-enabled format to interact with your simulated potential attendees and then an AI generated output to determine if what the attendees produce is what you want. This can be used for real-world events, but could theoretically be generalized to messy group collaborative AI agent creative interactions.

---

More Detailed AI-Generated Project Description:
The Velvet Rope reimagines event curation as a live, embodied AI experience where your face is the interface. Instead of a chat window, users encounter a six-stage theatrical pipeline: applicants are ingested from a CRM export, Claude generates dramatic first-person “pleas” for each candidate begging to get into an exclusive event, and a webcam-powered “Bouncer” stage reads the user’s facial expressions in real-time using face-api.js to compute a “Yes-ness” score — a weighted biometric signal derived from happiness, surprise, disgust, anger, and other micro-expressions detected at 4Hz. The user doesn’t click “approve” or “reject” — they simply react naturally to each pitch while OpenAI’s TTS HD engine (onyx voice) delivers the plea aloud with karaoke-style word highlighting. Their involuntary facial responses become the admission criteria.

Claude’s reasoning drives the entire generative pipeline: it writes individualized personality-aware pitch monologues calibrated to each applicant’s background, then powers a multi-round agent-based social simulation on “The Floor” where admitted attendees form groups, debate topics, and produce collaborative outputs. Claude determines who gravitates toward whom based on personality types, complementary skills, and shared interests — generating vivid per-round narratives, group cohesion scores, and emergent breakthroughs or conflicts. The venue visualization shows agents clustering in real-time with spring physics, attraction lines connecting group members and repulsion indicators between mismatched agents.

The prototype achieves stability through a fallback-first architecture: every AI-dependent stage (profile enrichment, pitch generation, biometric analysis, simulation) has deterministic fallback paths, so the full interactive experience works even without API connectivity. State is persisted to localStorage via a synchronous reducer, surviving page refreshes and enabling non-linear navigation with back buttons at every stage. The UI shatters conventions by replacing forms and buttons with biometric input, replacing chat with theatrical staging, and replacing static dashboards with a living social simulation — the interface itself is the product of Claude’s reasoning, not a wrapper around it.

We started from scratch.