Guides

AI Programmatic Advertising: How AI Buys Media, and How to Buy Media Inside AI

The short answer

AI programmatic advertising means AI that optimizes bids, audiences and creative inside the buying stack, and placements inside AI conversations bought programmatically. IAB Tech Lab's 2026 Agentic Roadmap extends OpenRTB and existing standards for AI agents through ARTF, Agentic Audiences and AAMP. AI-conversation inventory adds a real-time intent signal and a short decision window.

Thrad campaign composer with a campaign brief field, Add assets, Create manually and Draft with AI controls.

For a media buyer, "AI programmatic advertising" covers two separate changes. Software inside the buying stack now sets bids, chooses audiences and writes creative, and is starting to act as an agent that negotiates and executes the buy itself. Meanwhile the conversations people have with AI assistants are becoming ad inventory of their own. Most writing on the phrase covers only the first change.

What is AI programmatic advertising?

Programmatic advertising is automated, auction-based media buying, usually through a demand-side platform (DSP) bidding on inventory in real time. Add "AI" and the phrase can mean either of two things:

  • AI inside the stack: machine learning and large language models that optimize bids, targeting, creative and measurement on existing inventory such as display, video, CTV and audio.
  • AI as the inventory: placements inside the answers of AI assistants and chatbots, bought programmatically like any other channel.

IAB Tech Lab's AI in Advertising Primer frames the first meaning around the two activities it calls fundamental to advertising: "the scientific pursuit of optimization, and the art of creating persuasive media." The second meaning is newer, and it changes the buying mechanics more.

How is AI used inside programmatic buying?

Inside the stack, AI mostly automates work that used to be manual:

  • Bidding and pacing: models predict the value of each impression and set bids and budgets toward a goal.
  • Audiences and context: models infer who is likely to respond, and increasingly what a page or session is about, without relying on third-party cookies.
  • Creative: generative models produce and vary headlines, images and video for testing.
  • Operations and insight: assistants inside buying platforms recommend changes, resolve account issues and summarize performance. Meta's Ads Manager business assistant, covered in our Meta AI ads explainer, is one example.

Your ads still run where they ran before. What improves is the efficiency of buying the inventory you already buy.

What are the agentic advertising standards?

The larger structural change is agents: software that can plan, negotiate and execute a media buy, talking to other agents on the sell side. IAB Tech Lab, the industry's technical standards body, released an Agentic Roadmap on January 6, 2026. Its premise is to extend the standards that already run programmatic (OpenRTB, AdCOM, OpenDirect, VAST and the Deals API) rather than replace them, and to connect them to newer protocols including Model Context Protocol, Agent2Agent and gRPC.

StandardWhat it coversReleased
ARTFAgents inside real-time biddingNov 2025
Agentic AudiencesIntent signals between agentsNov 2025
AAMPBuyer and seller agentsJan 2026
CoMPPublishers and LLM contentWorking group
IAB Tech Lab's agentic and AI standards, as of October 2026.
  • Agentic RTB Framework (ARTF) defines how large language models and autonomous agents can take part in real-time ad transactions "without sacrificing performance."
  • Agentic Audiences is an open standard for how agents exchange identity, contextual and reinforcement signals that represent a consumer's real-time intent and response to advertising.
  • AAMP (Agentic Advertising Management Protocols) organizes the program into three pillars, foundations, protocols and trust, with reference buyer and seller agents published on GitHub. Its 3.0 release targets the RFP-to-buy process.
  • CoMP (Content Monetization Protocols) is a working group on how publishers and AI companies transact over content used by LLMs.

When the roadmap launched, IAB Tech Lab's CEO said: "Agentic execution is already part of how digital advertising operates today." For buyers that means continuity. The deal types, auctions and measurement you know are being extended, not thrown away.

What changes when AI conversations become inventory?

The buy itself looks familiar: an auction, a bid, a creative, a report. Three things are different:

  • The signal is the conversation. Instead of a cookie profile or a search keyword, the strongest signal is what the person is asking about right now. That is the kind of real-time intent signal the Agentic Audiences standard is designed to carry between systems.
  • The decision is made mid-answer. The ad has to be selected while the assistant is responding, so the decision window is very short.
  • The format is native. Placements appear inside or alongside an answer, so relevance and tone matter as much as reach.

Our guides to conversational advertising and advertising on ChatGPT show how these placements work in practice.

How do you buy programmatic ads inside AI conversations?

Thrad is a DSP for ads inside AI conversations: advertisers bid on conversational intent in real time, with delivery in under 500 milliseconds, across 20+ countries, with one budget and one report across every AI app. You can launch a campaign from self-serve or read how programmatic buying for the AI era works on Thrad.

Agencies and DSPs that want to bring this inventory into their existing buying setup can talk to our team about integrations.

Put conversational intent to work

Explore how to run ads inside AI conversations, choose a budget and measure the response.

Explore self-serve campaigns

What should buyers evaluate?

Ask these of any vendor, whether it sells AI tools for your existing stack or AI-conversation inventory:

  • Transparency: can you see which AI apps and surfaces your ads ran in?
  • Intent and safety: how are conversations classified, and how are sensitive or unsafe contexts kept out?
  • Speed: can decisions be made inside the response window without slowing the experience?
  • Measurement: what is reported (impressions, clicks, conversions), and can it reach your own analytics?
  • Reach: which markets and languages are covered today?
  • Standards: does it work with the programmatic standards your team already uses?

A checklist for AI programmatic buyers

  1. Separate the two budgets: AI tools that optimize existing channels, and new AI-conversation inventory.
  2. Ask every vendor where ads run, how intent is classified and what reporting you get.
  3. Follow IAB Tech Lab's agentic program; ARTF, Agentic Audiences and AAMP will shape how agents buy.
  4. Start AI-conversation inventory as its own line, with its own measurement, before folding it into always-on plans.
  5. Plan by market: AI-conversation inventory varies by country and language.

Common questions

What is AI programmatic advertising?

It can mean two things: using AI inside the programmatic buying stack to optimize bids, audiences, creative and measurement, and buying placements inside AI assistant conversations programmatically.

What is agentic advertising?

Media buying and selling carried out by AI agents that can plan, negotiate and execute transactions. IAB Tech Lab's 2026 Agentic Roadmap sets out how existing standards are being extended to support it.

What is the Agentic RTB Framework (ARTF)?

An IAB Tech Lab framework, released for public comment in November 2025, that defines how large language models and autonomous agents can take part in real-time advertising transactions without sacrificing performance.

Is OpenRTB being replaced by agentic standards?

No. IAB Tech Lab's roadmap extends established standards such as OpenRTB, AdCOM, OpenDirect, VAST and the Deals API with newer protocols, rather than replacing them.

What is an AI DSP?

A demand-side platform that either uses AI heavily to optimize buying, or buys AI-conversation inventory. Thrad is a DSP for ads inside AI conversations.

Can I buy AI-conversation inventory through my existing DSP?

Agencies and DSPs that want to bring Thrad's inventory into an existing buying setup can talk to our team about integrations.

Next step

Ready to test ads inside AI conversations?

What's the call about?