An agent for the people who fill seats

Your emptiest Tuesday,
programmed with taste.

Theme Night GM is an AI promotions agent for sports teams. It reads what your city actually loves through Qloo's taste graph: who those fans are, where they live and whether they're growing. Then it turns your weak home dates into theme nights with sponsors, a playlist, local partners and promo copy. Every pick comes with receipts.

Watch it plan Durham Brief it on your team

No sign-up. Replays a recorded live run instantly, or run it live (about two minutes).

What the GM weighs for one dateillustration

Tue

Apr

Night game · school night · vs Norfolk

Which crowd? Which fandom?

Local affinity

×0.3

Audience fit

×0.25

Fans near the venue

×0.2

Momentum

×0.15

New-fan reach

×0.1

Theme nights fill seats. They're planned on gut feel.

Minor-league and mid-market clubs lean on promotions to sell weeknight tickets: bobbleheads, fireworks, "Night" after "Night". A small promotions staff plans a whole season of them every off-season, usually from the same list every other club uses. Nothing in that process tells them what their market loves, whether those fans live near the ballpark, or which sponsor would pay to be part of it.

LLM alone

Suggests the same Star Wars / Harry Potter / 80s nights for every city
Can't tell you if the fans live 5 km or 50 km from your venue
Guesses which audience a fandom skews to
Invents sponsor fits with nothing to show a brand

Theme Night GM + Qloo

Finds fandoms your city over-indexes on: local rank vs national rank
Maps where those fans concentrate around your venue (urn:heatmap)
Scores each fandom for the date's crowd with demographic & life-stage signals
Pitches sponsors whose Qloo audience shares the taste, with numbers to quote

How the agent works

Eight tools, roughly 100 Qloo requests, one season plan.

A Gemini 3.8 Flash agent plans and calls tools. Each tool is a Qloo workflow that bundles the requests needed to answer one question a promotions director would ask. You can watch every step live.

Read the room

Asks Qloo what your city has the strongest affinity for across movies, TV, artists, video games and books, and where it over-indexes vs national popularity.

/v2/insightssignal.location.queryfilter.popularity.minbias.trends

Size the fandom

For each candidate: who the fans are, whether they live near your venue, whether interest is rising, and the taste tags that describe them.

urn:demographicsurn:heatmap/v2/trendingurn:tag

Fit the audience

Scores the shortlist for each date's crowd (families, Gen Z, young pros, 55+) and measures overlap with your existing fan base, to find nights that bring new people.

signal.demographics.agesignal.demographics.audiencesfilter.results.entities/v2/audiences

Build the night

Finds sponsor brands that share the fandom's taste, the artists for the in-game playlist, places within 6 km for pre-game partners, and podcasts for media buys.

urn:entity:brandurn:entity:artisturn:entity:place + filter.locationurn:entity:podcast/v2/analysis/compare

Ship it with receipts

The agent submits a plan, and the server rejects any entity Qloo didn't return. Each night gets a Taste Fit Score, a licensing flag and the exact requests behind it.

validated IDsTaste Fit Scorerequest logcopy-as-curl

Market DNADurham is not Los AngelesCompare any two markets side by side: what only one city loves, what both share, and how far each pick over-indexes locally. That local edge is what every night the GM plans is built on.Compare markets

Taste Fit Score

One number a GM can defend

Computed by code from Qloo responses, not written by the model, so it is consistent across runs and explainable line by line.

30%Local affinityRank percentile of this fandom among your city's top 25 in its domain, ranked by Qloo with the city as location signal (/v2/insights · signal.location.query).
25%Audience fitHalf: rank percentile when the same pool is re-ranked with the night's age or life-stage audience signal. Half: urn:demographics alignment for that segment.
20%Fans near the venueShare of the fandom's metro hotspots (top 20% of Qloo heatmap cells) inside the venue's 16 km catchment, relative to the catchment's share of all cells. 0.5 = fair share (urn:heatmap).
15%MomentumChange in Qloo trending percentile across a 16-week window (/v2/trending; uses the latest window with data).
10%New-fan reachInverse of how highly existing sport fans (a league proxy entity) rank this fandom: high = brings people who aren't coming yet.

Responsible by design

Aggregate taste, never people

No personal data goes to Qloo. Qloo only sees market-level inputs: the city, venue coordinates, entity IDs and age or life-stage signals. Results are aggregate audience affinities.
No sensitive targeting. Segments are age and life stage only. The validator also blocks identity-framed nights, sensitive anchors, and alcohol or bars on family and Gen Z nights.
Every claim has a receipt. Each night lists the exact Qloo requests behind it, and you can copy any of them as curl.
IP-aware. Studio and publisher metadata flags nights that need a license, and the license-free mode avoids them.
Can't cite made-up data. The plan validator rejects any entity ID that Qloo didn't return during the run.

Pick a market. Watch the GM work.

Durham baseball, LA baseball, Portland soccer, Nashville hockey, or your own team, city and schedule.

Open the studio