Advertising Events and Experiences Inside AI Conversations

Someone sorting out a weekend tells an assistant more in one message than they would ever type into a search box: the city, the dates, how many of them are going, what they are willing to spend. Thrad is now a distribution partner for Gomry, whose catalog covers live experiences and the tickets to them — so that kind of message can surface a specific show on a specific night instead of a shrug about what the city is like in October.

Three parties, three jobs, and it helps to keep them apart. Thrad picks the sponsored placement that fits the conversation. The AI publisher decides whether a placement is welcome there at all, renders it, and labels it. Gomry holds the catalog, the prices and the order. What follows is why that split exists and what each party is actually on the hook for.

Plans carry constraints a search query never states

A search query is two or three words and a lot of guesswork behind them. A planning message is closer to a brief. It usually names a place, a date range, a group size and a ceiling on spend, and it often throws in the preference that actually decides the outcome — outdoors, nothing that runs past eleven, walkable from where we are staying.

Those are the same fields an events catalog is already organised around. Date, venue, city, price band. Nothing has to be inferred from a click history, because the person has just said it out loud.

Timing matters more here than in most categories. An event is perishable in a way a pair of running shoes is not: it happens once, the seats run out, and a perfectly targeted placement that lands a fortnight after the gig is worth nothing at all. Conversations are where the deciding happens, which is why they are the right place for this inventory to show up.

What one message sets in motion

Take a request along these lines:

“Two of us are in Austin the second weekend of October. We want live music, outdoors if possible, and we would rather not go past $90 each.”

Four things happen, in order:

• The publisher hands over context. Its app is integrated with Thrad through an API or SDK, and the conversational context travels with the request.
• Thrad matches and ranks. Eligible sponsored placements are scored against what the conversation is actually about, inside whatever the publisher has allowed — which categories, which surfaces, how often.
• The publisher renders and discloses. If a Gomry experience wins, it appears in the publisher's own interface, in the publisher's own design, marked as sponsored.
• Gomry takes it from there. Follow the placement and you reach the catalog, where the price and the remaining availability are whatever they are at that moment rather than whatever they were when the placement was built.

Who decides what

Most confusion about this arrangement comes from assuming one party owns a decision that belongs to another. The short version:

Three things a placement is not

Worth being blunt about these early, because they are cheaper to explain before a campaign is in market than after.

• It is not a hold. Showing a placement reserves nothing. The tickets stay available to everyone else until Gomry returns a confirmed order.
• It is not guaranteed delivery. Sitting in the catalog earns an event the right to compete, not the right to appear. Relevance can rule it out, and so can the publisher.
• It is not a checkout. Whether someone can buy without leaving the assistant depends entirely on how far that app has taken its own commerce integration.

Where to start

If you run an AI product, the first move is integrating Thrad and then answering a product question rather than an ad-tech one: where does a real, dated, priced thing to do belong in your interface? Mid-answer, in a summary, off to the side. If the app is also going to handle the search and the purchase itself, Gomry's commerce documentation covers that half.

If you are buying, the same context matching is already running in categories with the same planning shape and in ecommerce. Events are a sharper version of the pattern, not an exception to it: the constraint set is tighter and the window closes faster.