Today, buying and selling digital advertising involves a lot of human coordination - emails, rate cards, negotiations, spreadsheets, campaign setup. A single campaign can take weeks to go from brief to live. AI agents are about to automate much of that. Here’s what that means, how far along it is, and what it means for your part of the industry.
IAB Tech Lab has just released AAMP 2.0, (Agentic Advertising Management Protocol 2.0) a set of tools that lets AI agents handle the negotiation and booking of digital advertising deals. Not as a concept or a pilot, but as working software that can be downloaded today.
What is AAMP 2.0?
Think of it as two AI agents - one representing the buyer, one representing the seller - that can find each other, negotiate a deal, and book a campaign without a human having to manage every step.
The Buyer Agent can take a campaign brief - budget, audience, KPIs - and automatically research inventory, allocate spend across channels, and generate a media plan. In AAMP version 1.0 this only worked for direct deals. Version 2.0 adds programmatic, which is where most of the volume is.
The Seller Agent turns a publisher’s media kit from a static PDF into an intelligent, dynamic storefront. It knows who the buyer is, what stage the deal is at, and adjusts pricing and packages accordingly.
The two agents speak the same language - built on existing industry standards like OpenDirect and OpenRTB - so they can transact with each other, or with traditional systems that haven’t gone agentic yet.
Humans in the loop
Humans are still in the loop - by design. Approval gates let people review and sign off at key points, and everything is logged and audited. The idea is not to remove humans from media buying, it’s to remove the less interesting, error-prone coordination work so that strategists can focus on strategy.
If I’m a programmatic media buyer in New Zealand, how would I use this?
As a programmatic buyer in New Zealand you’re probably using a DSP and doing a mix of open auction and private marketplace deals. The planning, prospecting, and deal negotiation around that is still largely manual - spreadsheets, emails, calls with publisher sales teams.
AAMP 2.0 doesn’t plug into your DSP tomorrow morning. It’s an SDK - a set of building blocks that DSPs, agencies, and ad tech vendors build with.
The realistic adoption path
Stage 1 - Your DSP or agency holding company builds on it
The first place you’d feel this is if your DSP or agency network integrates AAMP-compatible agents into their platforms. You might see it as a “campaign setup assistant” or an automated PMP deal discovery tool. GroupM, Publicis, and others are already experimenting with agentic buying infrastructure globally.
Stage 2 - Publishers you work with adopt the Seller Agent
If NZME, Stuff, or a local premium publisher stood up a Seller Agent, their inventory would become discoverable and negotiable by compatible Buyer Agents. Instead of emailing a sales rep for a rate card, an agent retrieves it automatically.
Stage 3 - You’re directing agents, not executing tasks
Instead of building a media plan manually, you brief an agent - audience, budget, KPIs, brand safety requirements - and it comes back with a plan, deal recommendations, and draft bookings for your approval. Your job shifts to evaluating and refining, not assembling.
What you’d actually do differently
- Less time on deal admin and campaign setup
- More time on strategy, creative direction, and performance analysis
- Approval and oversight rather than execution
- Probably need to get comfortable with how to brief an AI agent effectively - which is a new skill
The New Zealand market caveat
The NZ programmatic market is small compared to global markets. Global DSPs and publisher tech platforms prioritise features for US and UK scale. AAMP adoption here will lag the major markets most likely by about 12 to 24 months. The buyers who’ll feel it first are those working on global campaigns through international holding company networks.
However, understanding what AAMP is now means you won’t be caught off guard when your DSP rolls it out as a feature.
What this means for you
Publishers
Your media kit is about to become infrastructure. If your inventory isn’t structured, tagged, and machine-readable, AI buying agents simply won’t find it - or won’t prioritise it. The publishers who move early to make their inventory discoverable within agentic frameworks will have a significant advantage. For New Zealand publishers this means starting to think about how your inventory is described, packaged, and surfaced - not just to human sales teams, but to automated systems. The sales rep relationship doesn’t disappear, but it moves up the value chain toward strategy and partnership, away from rate card emails and availability checks.
Ad tech platforms
If you operate in New Zealand and your platform isn’t building toward AAMP compatibility, you risk becoming a gap in the supply chain. The standards being set now - how agents identify themselves, how deals are structured, how transactions are logged - are becoming the connective tissue of programmatic. Platforms that adopt early become easier to work with at scale. Those that don’t may find themselves excluded from agentic workflows by default.
Media agencies
This is the most significant structural shift for agencies. The value of an agency has always included the execution layer - building plans, negotiating deals, managing campaign setup. AAMP-style automation compresses that layer. The agencies that thrive will be those that reposition around what agents can’t do - judgment, relationships, creative strategy, and accountability for outcomes. Practically, this means starting to build internal capability around agentic tools now.
Clients and advertisers
In theory, agentic buying means faster campaign setup, less manual overhead, and more consistent execution against your brief. In practice, the near-term opportunity is better visibility and audit trails - knowing exactly how your budget was allocated, what deals were struck, and why. The question to start asking your agency or DSP is how they’re preparing for agentic workflows, and what human oversight looks like when AI is doing the negotiating on your behalf. It is important to remember that transparency and control matter more, not less, when automation is involved.
What’s next?
AAMP 3.0 is already in development, extending into more transaction and media types. The global trading infrastructure is being rebuilt around agentic AI. The standards being set now will shape how programmatic works for years, so now is a good time to start paying attention.
The SDKs, specifications, and tools referenced in this article are live and publicly available today. Here is what each part of the industry can do right now, with links to the relevant resources.
Publishers
The most useful thing New Zealand publishers can do right now is understand what’s coming. The Seller Agent SDK shows how agentic buyers will expect to discover and transact with publisher inventory in the future - and it reveals what machine-readable, well-structured inventory will need to look like. You don’t need to build anything today. But understanding the direction will help you make better decisions about how you package and describe your inventory going forward. If you have a technical team, the Agent Registry is also worth exploring - it’s a directory where publishers list themselves so automated buyer agents can find them as the ecosystem matures.
Seller Agent SDK (technical resource for developer teams): github.com/IABTechLab/seller-agent
Agent Registry (list your organisation for discovery by automated buyers): tools.iabtechlab.com/agent-registry
Agencies and media buyers
Review the Buyer Agent SDK to understand how agentic media planning and deal negotiation will work in practice. This is the clearest signal yet of where the buying process is heading.
Buyer Agent SDK (technical resource for developer teams): github.com/IABTechLab/buyer-agent
Ad tech platforms
The OpenRTB community extension and Prebid Agentic Audience Adapter are both live and available for adoption today. If you operate in the programmatic supply chain, these are the standards being built around you.
OpenRTB Agentic Audiences extension: github.com/InteractiveAdvertisingBureau/openrtb/blob/main/extensions/community_extensions/agentic-audiences.md
Prebid Agentic Audience Adapter: docs.prebid.org/dev-docs/modules/agenticAudienceAdapter.html
Open-source scorer: github.com/IABTechLab/agentic-audiences/tree/main/src/user-embedding-to-campaign-scoring
Read the full AAMP 2.0 article and the Agentic Audiences specification for the complete technical picture.
Full AAMP 2.0 article: iabtechlab.com/aamp-2-0-release-brings-transaction-ready-buyer-and-seller-agent-sdks
Agentic Audiences specification: iabtechlab.com/standards/agentic-audiences
