Seminole Hard Rock Watch

Watches the secondary ticket market for upcoming Hard Rock Live shows at a configured Seminole Hard Rock property. Uses an ai: arm to judge value against typical pricing for that venue rather than a flat threshold — because Hard Rock Live pricing varies wildly by act, but the venue's typical floor / mezz / balcony ratios are fairly stable.

What it does

On each fire:

1. The check calls a ticket-pricing endpoint for upcoming shows at the configured Seminole Hard Rock property (default hollywood-fl).
2. The ai: arm receives the listings JSON and a prompt that bakes in: - The venue identity ("Seminole Hard Rock Hollywood FL"). - That venue's typical price floors per section. - Concert-specific "good deal" criteria (buyer-regret drops, off-section listings, late-window supply-vs-demand).
3. If the AI flags one, notify with reasoning + full listings. Else stay quiet.

Why no season-ticket-holder dumping

Hard Rock Live is a concert/event venue, not a sports team. The "season-ticket holder selling individual games" pattern doesn't apply here — each show is a one-off. Late drops still happen, but for different reasons: - Initial buyers overpaying at on-sale, then trying to recoup. - Shows that didn't sell out the room having day-of marketplace supply. - Act-quality mismatches with venue size (smaller draw than expected).

This monitor is calibrated for those patterns rather than the sports-season-holder pattern in event-price-watch.

Which property are you watching?

Hard Rock Live operates at multiple Seminole Hard Rock properties. The venue= URL parameter in monitor.yaml's check.args.url selects which:

The AI prompt currently bakes in Hollywood FL specifics (capacity, typical price floors, "rock/country/comedy/Latin/R&B" act mix). For a different property, edit the prompt to match that property's:

• Venue capacity (changes "obscure act = upper balcony cheap" baseline).
• Typical act genres booked there.
• Per-section price floors — these vary by market. Tampa runs ~10-20% cheaper than Hollywood for comparable acts; Tulsa cheaper still.

Data-source setup

Same scraper / API endpoint pattern as event-price-watch. Your endpoint needs to accept a venue filter and return upcoming-show listings within a windowDays range.

Expected published bindings from the scraper:

A minimal listing record:

{
  "showName": "<artist>",
  "showDate": "2026-06-15",
  "section": "Floor B",
  "row": "12",
  "price": 85,
  "note": "",
  "listingUrl": "https://marketplace.example.com/listing/abc123"
}

The note field is where obstruction tags, alcohol-included, or "VIP package" extras go — surface those to the AI so it can disqualify or correctly value them.

Edit before installing

In monitor.yaml:

• check.args.url — set venue= to your property and point at your scraper.
• The ai: prompt — adjust the venue name, capacity context, and typical price floors to match the property you picked (see table above for which Seminole properties have Hard Rock Live).
• when[0].do[0].subject / body — adjust copy.
• Schedule: daily is plenty. Concert listings move slowly until the final 72 hours; once a day catches the long-tail drops without burning LLM budget.

Bindings used

• {{check.listingsJson}} — JSON array of upcoming-show listings.
• {{ai_reasoning}} — AI's explanation when it flags a listing.
• {{monitor.firedAt}} — fire timestamp.

Tools used

• http — core InTouch HTTP task type.
• AI assistant (Anthropic / OpenAI / Gemini / Ollama).

Pattern showcase

Venue-aware AI valuation — the AI prompt encodes the specific venue's pricing model, not a generic one. This is the right shape any time the "good deal" definition depends on knowledge that's static per-installation but varies across users (different venue, different stock portfolio, different industry). Bake the per-venue knowledge into the prompt at install time; the data source stays generic.

Related monitors

• good-seat-ai-watch — sports-event variant (lower bowl / mezz / upper bowl floors instead of Hard Rock Live's floor / loge / balcony).
• event-price-watch — single-event raw threshold, no AI.
• gameday-firesale-watch — last-hours panic-drop detection.