Agent-to-agent
A brand's buying agent transacts directly with a surface owner's sales agent. No bidder skimming the middle, no daisy-chained intermediaries.
Agentic advertising.
Agentic advertising is a new way to advertise. AI agents act on behalf of brands, transacting directly with agents from surface owners — and carrying brand strategy, standards, and budget into every decision across every surface where consumers are.
We have been building this alongside some of the most forward-thinking brands in the world.
Definition
What is agentic advertising?
Agentic advertising is a model where AI agents — buying agents on behalf of brands and sales agents on behalf of surface owners — discover, negotiate, and transact across every surface where consumers are. The buying agent carries the brand's objectives, standards, and budget. The sales agent represents the surface owner's inventory and terms. The two communicate directly through an open protocol called AdCP. The result is a faster, more transparent, more flexible advertising market — one where brands lead and the system follows.
AI agents are already buying advertising. The brands moving first will define how it works.
The world's largest brands are putting agents in front of real budgets. The infrastructure is forming, the protocols are open, and the consumer behavior is in motion.
66%
of ad buyers are increasing focus on agentic AI for ad buying and campaign execution, with 96% already aware of it as a capability.
IAB
60%
of brands will use agentic AI to deliver one-to-one marketing interactions by 2028 — what Gartner calls "the end of channel-based marketing as we know it."
Gartner
75%
of consumers are open to using an AI-powered personal shopper that understands their needs and preferences.
Accenture
Agentic advertising vs. programmatic advertising.
Programmatic was a real-time bidding solution to a 2010 problem: how to automate the purchase of standardized digital impressions. Agentic advertising solves a 2026 problem: how to coordinate strategy, surfaces, and commerce across an exploding map of formats — including ones that auctions were never designed to support.
| Programmatic advertising | Agentic advertising | |
|---|---|---|
| Buying decision | Real-time bidding on individual impressions. | Strategic allocation across a portfolio of surfaces and outcomes. |
| Decision logic | Pre-configured rules, auction mechanics, opaque optimizers. | Brand objectives, standards, and judgment translated by the agent. |
| Counterparty | A chain of intermediaries between brand and surface owner. | Buying agent transacts directly with the surface owner's sales agent. |
| Surfaces reachable | Whatever fits the auction pipe. | Any surface a sales agent represents — including non-programmatic. |
| Governance | Bolted on after the fact. | Structural — applied to every transaction before it executes. |
| Economics | Take-rate stacks across the supply chain. | Transparent fees. More of every dollar reaches the audience. |
"The way RTB unlocked programmatic, AdCP unlocks agent-to-agent advertising."
How agentic advertising works.
A brand expresses goals, audiences, offers, and standards once. The buying agent carries that context into every decision. Sales agents on the other side of the marketplace represent surface-owner inventory directly. The two transact through AdCP.
- 1
Brief
The brand describes the campaign in natural language — objectives, audiences, budget, standards, governance. Once.
- 2
Discovery
The buying agent queries sales agents across the marketplace through AdCP. It evaluates surfaces, audiences, formats, and pricing simultaneously, surfacing what fits the objective.
- 3
Negotiation
Agents agree on terms — pricing, targeting, delivery, creative requirements — within the boundaries the brand has set.
- 4
Execution
Brand standards, audience guardrails, and approval thresholds are checked on every transaction before it executes. Nothing runs outside those boundaries.
- 5
Reporting and learning
Outcomes flow back through the same protocol. Every transaction is logged in full and feeds the next brief. The agent that has been running for a year is working with knowledge the brand spent a year accumulating.
Every surface, on the brand's terms.
Consumer attention has never been more distributed. A morning commute touches audio, transit screens, and a dozen social feeds. An evening at home moves between streaming, social, AI assistants, and retail environments. The line between content, commerce, and customer relationship is dissolving — a consumer asking an AI assistant for a product recommendation, discovering a brand inside a retail surface, and completing a purchase in the same session.
Surface owners build sales agents that represent everything they sell, on their own terms. Every format, every audience, every context — listed once and discoverable to every brand whose buying agent speaks AdCP. A podcast network, an AI assistant, a retail surface, a sponsorship opportunity: all within reach.
The brand's agents adapt to each surface's strengths while remaining distinctly the brand. A luxury brand does not show up the same way on a sports podcast as it does on a premium streaming platform — but it shows up as itself on both.
A non-exhaustive list
Social
Branded content, influencer placements, sponsored posts.
Audio & podcasts
Host-read, dynamic insertion, programmatic audio.
CTV & streaming
Pre-roll, mid-roll, sponsorships across streaming platforms.
AI assistants
Sponsored Intelligence inside ChatGPT, Claude, Copilot, and beyond.
Retail media
On-platform placements across retail networks.
Display & rich media
Standard and high-impact formats across web and app.
Out-of-home
Digital billboards, transit, place-based formats.
Newsletters & email
Sponsored placements, native units, dedicated sends.
Events & sponsorships
Conference, webinar, experiential packages.
In-game & live
Custom in-game experiences and AI-delivered placements within live broadcasts.
Data products
First-party segments, contextual data, research panels.
And surfaces yet to come
Every new surface that adopts AdCP becomes reachable to every buying agent.
Sponsored Intelligence: brands inside the moment of decision.
AI is becoming the primary interface where people discover, research, and decide. The conversations happening inside assistants, copilots, and agentic experiences are advertising surfaces in their own right. Sponsored Intelligence is the economic model that lets brands participate in those moments.
In every prior advertising model, ads were delivered adjacent to content — the banner next to the article, the pre-roll before the video. Sponsored Intelligence narrows that gap. Even where the placement appears alongside a response, it is surfaced inside an active decision-making moment and matched to explicit intent in real time.
OpenAI is running ads inside ChatGPT. Amazon is exploring infrastructure for AI assistants on third-party apps. Booking.com, Instacart, and Shopify are connecting commerce directly into chat. AdCP 3.0 formalizes Sponsored Intelligence as a named media surface, with standards for session management, capability negotiation, brand safety, disclosure, and commerce handoff.
What changes for brands.
The buying agents exist solely to serve the brand. Surface owners represent their inventory directly, without an intermediary deciding what gets seen or recommended. Economics are transparent. Every recommendation reflects fit for the brand.
Strategy, not bids
Decisions are made against brand objectives — not auction mechanics. The agent allocates budget like a portfolio manager, not a bid engine.
Every surface
Social, audio, AI assistants, sponsored intelligence, CTV, retail media, traditional media, out-of-home — and surfaces that have not been built yet.
Open by default
AdCP is the open protocol. Any buying agent can transact with any sales agent that speaks it. The marketplace stays open and the brand is never locked in.
Governance built in
Brand standards, placement rules, audience guardrails, and approval thresholds are enforced on every transaction — set once, applied without exception.
Improves with use
Every campaign teaches the agent something. The brand running agents for a year is working with agents that know its DNA.
Each campaign raises the floor for the next.
Every campaign teaches the agent something. A seasonal push across fifteen markets generates data on which surfaces performed, which audiences responded, which formats carried the brand most effectively. That knowledge informs the next brief, sharpens the next set of recommendations, and raises the floor of what the brand can expect.
A brand that has been running agents for a year is working with agents that know its DNA — its audiences, its standards, its patterns of performance across markets and surfaces. The brands building this practice now will carry an advantage that grows the longer they run.
In practice
After twelve months, the agent has learned that a particular demographic responds consistently better on audio surfaces in the morning and AI assistants in the evening for this brand's category. That insight informs the next brief automatically.
Governance architecture, built in from the ground up.
When agents act on the brands' behalf, governance is the framework that defines what they can do, how much they decide independently, and how the brand's judgment guides every decision. Brands set the policies, standards, and boundaries every agent operates within. Those rules are set once and enforced on every transaction, without exception.
A global brand managing dozens of sub-brands across sixty markets might require that no placement runs adjacent to content flagged for specific sensitivity categories, that spend in any single market never exceeds a defined threshold without human approval, and that new surface types are reviewed before activation. Each brand and market operates under its own configuration, maintained independently.
The level of automation is a brand decision, not a platform default — observation mode, guided mode, or autonomous within authorized boundaries. Switch modes at any time, for any brand or market. Every transaction is logged in full and exportable to compliance, finance, and measurement systems.
Brand standards
Placement restrictions, audience guardrails, budget limits, brand-safety thresholds. Set per brand, per market, per surface.
Approval workflows
Gate the decisions that matter most — high-spend buys, new surface activations, creative changes — behind human approval.
Human in the loop
Start with full human approval. Move to defined guardrails. Scale to autonomous execution as confidence grows.
Audit trail
Every decision logged with full context — what was bought, where, at what price, and under what authorization. Read-only access for compliance teams.
Compliance and privacy
GDPR and CCPA compliance built into the architecture. First-party data stays the brand's. Clear data-processing agreements available.
AdCP, the open standard that makes it possible.
For agents to buy advertising across every surface, they need a shared language — an open standard that any buying agent and any sales agent can speak. AdCP, the Advertising Context Protocol, is that standard. Built specifically for advertising, AdCP defines how inventory, formats, audiences, pricing, and terms are exchanged.
AdCP is published openly at agenticadvertising.org, not proprietary to any vendor. Scope3 is a founding contributor and built entirely on AdCP. That is why the marketplace keeps growing, and why brands that connect through Scope3 are not dependent on any single platform's decisions about what inventory they can access.
The recent AdCP 3.0 release adds a Sponsored Intelligence Protocol covering session management, capability negotiation, brand safety, disclosure, and commerce handoff — formalizing AI-native surfaces as a named media channel.
Works with what you already have.
For brands with existing AI infrastructure
Brands with existing AI investments connect through AdCP without rebuilding what they have already built. Brand agents, insight engines, and data infrastructure developed elsewhere connect to the platform through standard interfaces.
For brands working with agencies
The platform works within existing agency structures. How a brand collaborates with its holding company, specialist partners, and internal teams remains the brand's decision. Scope3 adds a capability without changing the relationships around it.
For brands managing a portfolio
Each brand and market operates under its own rules, standards, and approvals. The platform fits the organization, however the brands' teams and partners are structured.
Common questions about agentic advertising.
Agentic advertising is a model where AI agents act on behalf of brands and surface owners. A brand's buying agent carries the brand's objectives, standards, and budget. A surface owner's sales agent represents inventory, formats, audiences, and pricing. The two transact directly across every surface — social, audio, AI assistants, sponsored intelligence, CTV, retail media, traditional media, out-of-home, and surfaces that have not been built yet.
Programmatic advertising is real-time bidding on individual impressions through a chain of intermediaries. Agentic advertising is agent-to-agent: a buying agent transacts directly with a sales agent, decisions are made against the brand's objectives rather than auction mechanics, and the surfaces reachable are not limited to whatever fits the auction pipe. Where programmatic is tactical and reactive, agentic advertising is strategic and proactive.
A brand agent — also called a buying agent — is an AI buyer that acts on behalf of a brand. It translates objectives, budgets, and brand standards into decisions, and transacts directly with surface-owner sales agents. The brand stays in command of the strategy and standards the agent operates under.
Agent-to-agent advertising is the transaction model behind agentic advertising. The brand's buying agent and the surface owner's sales agent communicate directly — discovering inventory, negotiating terms, executing the buy, and reporting outcomes. There is no black-box bidder in the middle, and no take-rate stacking across intermediaries.
AdCP, the Advertising Context Protocol, is an open standard that lets any buying agent transact with any surface owner's sales agent. It defines how inventory, formats, audiences, pricing, and terms are exchanged. AdCP is published openly at agenticadvertising.org, not proprietary to any vendor. Scope3 is a founding contributor and built entirely on AdCP.
Any surface a sales agent represents — social, audio, AI chat and assistants, sponsored intelligence, CTV, retail media, traditional media, out-of-home, events and sponsorships, and surfaces that have not been built yet. Because AdCP is open, every new surface that adopts it becomes reachable to every buying agent that speaks the protocol.
Sponsored Intelligence is the economic model for advertising inside AI-native surfaces — assistants, copilots, and agentic experiences where consumers research and decide. Brands participate in the same reasoning context as the AI's response, matched to explicit intent in real time. AdCP 3.0 formalizes Sponsored Intelligence as a named media surface with standards for session management, capability negotiation, brand safety, disclosure, and commerce handoff.
Yes. Brand standards, placement restrictions, audience guardrails, budget caps, and brand-safety thresholds are set by the brand and enforced on every transaction before it executes — in every market, every language, and every regulated category. Every decision is logged in full and exportable to compliance, finance, and measurement systems.
Governance is structural, not bolted on. The brand defines the boundaries every agent operates within and the approval thresholds that trigger human review — per brand, per market, per surface, per spend tier. The level of automation is a brand decision: observation mode, guided mode, or autonomous within authorized boundaries. Switch modes at any time, for any brand or market.
Either. Brands with existing AI infrastructure connect their own agents through AdCP. Brands without one can use agents built on Scope3. Either way, the brand stays in command of the objectives, governance, and standards the agent operates under. Plug in your existing agent or use ours — no rebuild.
Agencies still matter. A holding company or specialist partner can operate Scope3 on the brands' behalf, or run their own agents alongside the brands'. Scope3 adds a capability without changing the agency relationships around it.
Scope3 takes a transparent margin on media that flows through the platform, built into media cost rather than added on top. Routed transactions are 4%; decisioned transactions are 8%. No hidden intermediaries, no daisy chains — more of every dollar reaches the audience. Volume discounts apply on monthly spend.
Create a Scope3 account, then connect it to the AI assistant you already use — Claude, ChatGPT, or Copilot — through the Scope3 connector. Start in observation mode, where the agent analyzes and recommends and a planner approves, and widen autonomy as confidence grows. Most brands are running their first campaigns within a day.
Keep going.
The Brand Platform
The platform, the marketplace, and how it works in practice for brands and agencies.
Storefronts for surface owners
How surface owners merchandise, sell, and scale every ad opportunity through agent-facing storefronts.
The Scope3 manifesto
The end of the internet as we built it — and why agentic advertising changes the math.
AdCP, the open protocol
The Advertising Context Protocol — discovery, negotiation, transaction, reporting. No lock-in.
Pricing
Transparent fees, tied to media spend. No hidden intermediaries.
Install Scope3
Connect Scope3 to Claude, ChatGPT, or Copilot in two steps. Most brands run their first campaigns within a day.