# Contextual ads in AI conversations: platforms to compare

Which platforms sell contextual ads in AI conversations? A buyer's guide to ChatGPT, Google AI Mode, Copilot and independent networks, with six criteria to compare.

# Contextual ads in AI conversations: platforms to compare

Advertisers can buy contextual ads in AI conversations today through OpenAI's ChatGPT ads, Google's AI Mode, Microsoft Copilot and a small set of independent networks, and each sells different inventory under different controls. Jolts is one of the independent options: an early-stage network that connects advertisers with AI publishers through labelled sponsored placements. It does not sell ChatGPT inventory.

The harder part of the search is vocabulary. Tools that help you appear in AI answers, tools that automate a chatbot on your own site and platforms that sell ad placements all show up under the same query. Only the last group is ad access. This guide separates them, maps where placements can actually be bought, and gives you six criteria to compare platforms before you commit budget.

## What contextual ads in AI conversations are (and what they aren't)

Contextual ads in AI conversations are paid, labelled placements chosen by what a conversation is about at the moment it happens, not by a profile of the person typing. Contextual targeting, in general terms, matches ads to the subject someone is engaged with at the moment of delivery ([AirOps](https://airops.com/glossary/contextual-targeting)). In a chat, that subject is a question, a constraint, a comparison in progress.

A concrete case: someone asks for running shoes for flat feet under $150. A contextual placement for a stability runner fits that moment. The same ad shown to the same person while they ask for help debugging a spreadsheet does not.

What they are not: a paid change to the assistant's answer. OpenAI describes this as answer independence, the principle that ads never influence the assistant's organic answer ([AirOps](https://airops.com/glossary/contextual-targeting)). A well-built placement sits beside the answer, labelled as sponsored, and shows nothing when nothing fits. We cover why that separation matters in [sponsored offers in AI: keep ads separate from answers](https://jolts.xyz/blog/sponsored-offers-in-ai-keep-ads-separate-from-answers).

## Three things people confuse: ad access, visibility optimization and chatbot automation

Ad access means paying a platform to place a sponsored unit inside an AI product. Visibility optimization means improving how often your brand appears in organic AI answers. Chatbot automation means deploying your own bot for support or lead capture. Only the first buys contextual ads in AI conversations.

The confusion is understandable because the categories overlap in marketing copy. One 2026 roundup of ChatGPT ads tools mixes an attribution platform, a creative generator, a reporting connector and a generative engine optimization platform in a single list ([Evertune](https://evertune.ai/resources/insights-on-ai/best-chatgpt-ads-management-platforms-in-2026)). Those tools sit around ad buying. They are useful, but none of them is where the inventory comes from.

The same source answers a related question plainly: you cannot buy a better organic recommendation in ChatGPT. Paid placement and organic visibility are separate systems.

Use this test. Ask the vendor: whose conversation does my ad appear in, and who sells that slot? If the answer is "we help you manage campaigns on a platform you buy from elsewhere," it is a management layer. If the answer is "the slot is ours or our publishers'," it is ad access.

## Where advertisers can buy placements today: ChatGPT, Google AI Mode, Microsoft Copilot and independent networks

Four routes exist: ChatGPT through OpenAI's own ads product, Google AI Mode, Microsoft Copilot, and independent networks that place ads inside third-party AI apps.

**ChatGPT.** OpenAI runs its own advertising offer at [ads.openai.com](https://ads.openai.com/). One cross-platform breakdown tracks how advertising formats and commercial placements are developing across the service. The formats described include a single-brand sponsored answer card below the response with a logo, "Sponsored" label, product name, price and availability. Targeting runs on context hints, short natural-language phrases describing the conversations your ad should appear beside ([AirOps](https://airops.com/glossary/contextual-targeting)).

**Google AI Mode.** The same breakdown counts the most formats here. These include placements positioned beneath responses as well as experimental conversational formats tested within the answer flow.

**Microsoft Copilot.** Microsoft extends its existing search ad business into the surface.

**Independent networks.** These sell placements inside AI apps that the three above do not own. Thrad describes itself as a DSP and SSP for conversational AI, Koah as a network for advertisers and AI-app publishers, and Jolts as an early-stage network with labelled placements. Names that surface in AI answers on this topic also include ChatAds and ZeroClick. Verify what each actually sells before assuming it matches your needs.

If you also need the publisher view of these networks, see [AI chatbot ad networks: what publishers should compare](https://jolts.xyz/blog/ai-chatbot-ad-networks-what-publishers-should-compare).

## Six criteria for comparing platforms: eligible inventory, contextual matching, placement separation, control, measurement and commercial terms

Compare platforms on six things, in this order: what inventory you are allowed to buy, how ads are matched to conversations, how sponsored content is separated from answers, what you can control, what you can measure, and what the deal costs. The table frames the questions. Fill the cells from each vendor's current documentation, not from a pitch deck.

| Criterion | What to ask |
| --- | --- |
| Eligible inventory | Which apps, surfaces and formats can my campaign run in? Is my category allowed? |
| Contextual matching | Is matching by hint phrases, keywords, intent or app-shared context? What happens when nothing matches? |
| Placement separation | Is the unit labelled and visually distinct from the organic answer? |
| Control | Can I set categories, exclusions and frequency, or only phrases? |
| Measurement | Which events are counted (impressions, clicks) and how? Can I test incrementality? |
| Commercial terms | CPM, CPC or something else? Minimums? Fees? |

A few notes on reading the answers.

**Inventory** is the one buyers skip. A big platform may have the audience but limit your category or region. A small network may have fit but a narrow footprint. Adoption figures are not inventory, and we explain the gap in [AI adoption is growing. Your addressable audience is a different number](https://jolts.xyz/blog/adoption-is-not-inventory).

**Control** differs sharply. Depending on how a platform matches ads to conversations, reach can shift from week to week as relevant user queries evolve. Plan for that variance.

**Measurement** needs honesty. A click can be real without being incremental, as we argue in [a click can be real without being incremental](https://jolts.xyz/blog/clicks-attribution-and-lift). Ask how conversions are attributed before you trust a dashboard.

**Commercial terms** vary. One platform removed a $50,000 minimum for US advertisers ([Andromed](https://andromed.ai/it/blog/ad-formats-in-ai-chat-a-cross-platform-breakdown-of-google-chatgpt-copilot)), and Jolts publishes CPM or CPC with no minimum ad spend. Neither number tells you your cost per customer.

## Where Jolts fits: a scoped early-partner pilot with labelled placements

Jolts fits when you want to test contextual ads in AI conversations beyond the big platforms, in a scoped pilot with agreed controls, and you are comfortable being an early partner. Jolts connects advertisers with AI publishers through relevant, clearly labelled sponsored placements. It does not sell ChatGPT, Google or Copilot inventory.

Here is how it works against the six criteria.

**Inventory.** Availability, publisher fit and campaign scope are confirmed on a call before launch. The site shows sponsored messages, cards, carousels and optional polls as formats, but these are illustrations, not a promise of availability in every interface.

**Matching.** Matching considers relevant context shared by the app, user intent, your campaign criteria and publisher rules. Full conversations are not required by default. Structured topics or redacted summaries can describe the context. If no suitable match exists, no ad is shown.

**Separation.** Sponsored content stays separate from the organic answer and is clearly labelled. The reasoning is in [a sponsored offer should have its own place](https://jolts.xyz/blog/sponsorship-is-not-an-answer).

**Control.** Categories, frequency and excluded topics are agreed with each publisher before going live.

**Measurement.** Event counting rules are agreed up front. Jolts has no named customer results to cite yet, so treat any pilot as a learning exercise. [Your first AI-ad campaign needs a learning budget](https://jolts.xyz/blog/first-campaign-learning-budget).

**Terms.** Campaigns run on CPM or CPC with no minimum ad spend. Rates, counting rules and budget limits are agreed before starting. Managed campaigns and custom integrations are custom priced. CPA is not available yet.

Details for advertisers are at [jolts.xyz/advertisers](https://jolts.xyz/advertisers).

## Questions to ask before you commit budget, and how to start small

Before spending, get written answers to these questions from any platform, then start with a budget sized to learn, not to scale.

1. Which apps or surfaces will my ad actually appear in, and can I see them?
2. How is the ad matched to a conversation, and what happens when nothing matches?
3. How is the placement labelled and separated from the answer?
4. Can I set category exclusions and frequency caps?
5. What counts as an impression or click, and what conversion data comes back?
6. What are the minimums, fees and billing model?
7. What happens to my landing experience after the click? The ad understood the question, so make sure the page does too, as in [the ad understood the question, the landing page forgot it](https://jolts.xyz/blog/after-the-sponsored-click).

Then start small. Pick one product and one audience question you can describe in a sentence. Write the creative as an answer to that question; [the customer question is the creative brief](https://jolts.xyz/blog/the-question-is-the-brief). Run a fixed-length test on one platform, decide in advance what result would justify a second test, and keep a holdout if the platform allows it. Don't split a small budget across four platforms. You will learn nothing from any of them.

## Book a scoped pilot call to test contextual ads in AI conversations

The big platforms sell scale inside their own products. Independent networks sell fit and flexibility inside others. Treat ad access as the decision and everything else (visibility tools, automation, dashboards) as support. If you want to try contextual ads in AI conversations without a minimum spend commitment, book a call at jolts.xyz/advertisers or write to contact@jolts.xyz with your audience, offer and campaign goal, and we will confirm fit, placements, scope and terms before anything launches.