Automation Disclosures

Gemini receives submitted flyer image content for structured event-detail extraction and public event/artist context for descriptions, music classification and artist enrichment. Inputs include event/artist names, dates, venues, genres, descriptions, biographies and images; public content can contain personal/contact information. Outputs affect catalog descriptions and classification. Organizer extraction is a draft for review; enrichment is not universally reviewed by a human before display. Verify factual event details with organizers and request corrections through support/claims.

When Replicate credentials are configured, event generation sends an organizer’s prompt to a hosted language model to draft event descriptions, suggested details and promotional copy; deterministic templates are used otherwise. Prompts can contain whatever personal details the organizer supplies. Do not include unnecessary confidential/personal information. Organizers must review bookings, prices, restrictions and claims before publishing. The image-generation endpoint is explicitly disabled; no current live image-generation service is claimed.

Internal statistical/behavioral models use music preferences, Spotify signals, RSVPs, follows, saves/swipes, purchases and city preferences to rank discovery and suggest follows. These modules do not by themselves send private taste profiles to a third-party LLM. Gemini also assists organizer outreach drafting from public organizer/event context; internal review and sending paths must be audited per deployment. SEO copy generation uses Gemini public topic/event context and is saved as an unreviewed draft until an admin approves it. No production AI support chatbot or LLM moderation service was verified.

Automated output can misstate dates, affiliations, artists or suitability, so confirm important details and report errors. Spotify disconnect is available, but a global personalization opt-out and complete derived-signal deletion were not verified. Provider input retention/training settings and processor contracts require review; we do not promise zero provider retention or no model training without verified settings. Claude/Codex development tooling is excluded from these product disclosures.