Marta KravsDirect deals are commercially valuable, but moving them from request to launch remains operationally heavy and highly manual. Prospecting, proposal alignment, deal setup, creative approval, and post-sale operations all require ongoing manual coordination.
Sales agents are gaining relevance as the buy-side adopts agentic AI to search for opportunities, assess fit, and move through workflows faster. As buyers move in that direction, publishers need a structured way to respond within the same workflows or risk losing visibility in those buying paths.
This is where AdCP sales agents step in. They automate coordination across direct deal workflows without removing publisher control. The sell-side still retains control over inventory, creative approvals, and deal-level decision-making. The sales agent automates the operational work around those decisions, including request matching, proposal routing, and deal progression through review.
As Anastasia-Nikita Bansal, our CEO, notes, market uncertainty and the novelty of the concept still make some publishers cautious about embracing agentic advertising. Behind all the terminology, some publishers still question how AI agents work in practice and what this model would actually change for their direct sales and PMP workflows. This article explains where those gaps still exist and how publishers can evaluate them more practically.
Limited ecosystem clarity slows early adoption
Anastasia-Nikita Bansal has pointed to a lack of operational clarity as a major reason many publishers hesitate to adopt the model. They may be curious, but without a clear picture of how to operationalize the agents, early-stage testing and the integration of AdCP agents into the tech stack stall.
As the market works through the initial wave of attention around AdCP sales agents, three testing barriers stand out:
Technical misunderstanding. Key workflow questions still come up repeatedly: who initiates the request, what information it contains, and who is expected to respond. Publishers also ask where verification happens, who handles approval, and where execution actually takes place.
Ecosystem misunderstanding. Many publishers still lack clarity on how buyer and publisher sales agents fit together. They also do not fully understand what exactly moves through the workflow: campaign briefs, inventory data, offers, approvals, and creative assets.
Operational ambiguity. Many publishers still question the value of AdCP sales agents if direct-deal relationships and basic negotiations are still necessary. They also want to understand which types of direct deals this model can realistically support, and whether using a sales agent requires separate internal ownership or a dedicated team.
AdCP sales agents make direct deals easier to scale
More than 90% of total US programmatic display ad spending flows through programmatic direct, and a large share of that activity still depends on manual coordination. Even in programmatic direct, software automates delivery and in-platform execution. But negotiation, approvals, and ongoing coordination still require human input across execution and post-sale operations.
Some steps alone can take multiple days. That includes waiting on buyer feedback, reconciling deal terms, collecting creative assets, and securing final approval on the publisher side.
Manual input implies greater control, but the downside is an operationally heavy process that consumes team bandwidth and slows down execution.
That is why publishers already working with direct deals stand to gain the most from AdCP sales agents. Here, scale does not automatically generate new demand. It means reducing the amount of manual work required for each deal. By standardizing inventory exposure, proposal exchange, approval routing, and the movement of deal information through the workflow, a sales agent can help the same team handle more direct-deal activity without adding headcount.
How an AdCP sales agent works without replacing the stack
An AdCP sales agent is publisher-side AI software built on AdCP, an open-source standard for agent communication. It does not replace the publisher’s stack. It works on top of it by making inventory and sales capabilities available in a machine-readable way.
The process starts with setup. The sales agent exposes the publisher’s available offerings through a machine-readable directory and connects to the existing ad server or sales infrastructure to retrieve product information and availability. That gives buyer agents a structured way to understand what inventory is available and under what terms.
From there, the sales agent supports discovery, negotiation, and workflow coordination. A buyer agent sends campaign information and requests matching inventory from the sales agent, either directly or through an activation platform where agent interaction occurs.
The sales agent then returns structured inventory and product details so the buyer-side system can evaluate fit and generate an offer. If adjustments are needed, the proposal can be refined within the same workflow rather than restarted from scratch.
If the buyer-side criteria match, the offer is routed to the relevant publisher sales agent for review. Creative approval still happens on the sell side, either by humans or through the sales agent, where rules allow.
Once both sides have aligned, the sales agent can formalize the agreement in a machine-readable format within the publisher’s existing system. If creative assets are part of the workflow, the agent can also support creative handoff and validation against format requirements before serving.
None of that removes governance or publisher control. The sales agent can run rule-based checks before a deal moves forward, including policy, brand safety, and compliance checks as needed. And if a deal falls outside pre-set authority limits, it can pause the workflow and escalate to a human decision-maker for approval.
Once approved, the campaign still runs through the existing delivery infrastructure. Ad serving continues via the seller’s own stack or via RTB-compatible infrastructure, such as header bidding.
How publishers can use AdCP sales agents to negotiate deals
Agentic advertising adds a strategic layer above RTB execution
A recurring question in adtech is whether agentic advertising could replace RTB. But the more accurate distinction is not agentic versus programmatic. It is about strategy and allocation versus execution.
Agentic systems are increasingly described as a layer above programmatic execution, where buyer and seller agents can reason about fit, allocation, and workflow decisions. The underlying transaction, however, still runs through deterministic infrastructure, such as auctions, header-bidding wrappers, and OpenRTB-compatible systems. In that sense, AdCP is better understood as an interface or allocation layer, not as a replacement for RTB execution.
That distinction matters because it clarifies what agentic advertising is actually meant to do. It does not replace the auction mechanics that already execute media trades efficiently. It adds a machine-readable layer for reasoning, coordination, and decision support around those mechanics.
AdCP sales agents shorten direct-deal workflows and reduce coordination overhead
For publishers already running direct deals, the main value is operational: less manual coordination across the workflow.
Practical testing offers the clearest view, but the benefits observed so far include:
Lower operational overhead through less manual coordination across requests, negotiations, approvals, and follow-up.
Faster handling of requests, proposals, and approvals because information moves through a more structured workflow with fewer manual handoffs.
More standardized deal coordination, with less variation in how requests, approvals, and communication are handled across the workflow.
Early results from TeqBlaze’s test group suggest that the agent can reduce time spent on requests, approvals, and other coordination-heavy steps. Based on our testing observations, the path from buyer outreach to ad launch can take one to two weeks. Under favorable conditions, where deal terms are aligned quickly, creatives are provided on time, and approval flows are straightforward, a sales agent can reduce that timeline to as little as 48 hours.
If additional testing confirms a strong operational fit, AdCP sales agents may help publishers process more direct deals without expanding the team. That becomes possible by reducing time spent on repetitive negotiation, contract handling, and approval workflows. As a result, sales and ad ops teams would have more time to pursue new partnership opportunities, explore additional demand paths, and support a higher volume of premium deals.
Automation helps, but publishers still own the critical calls
The biggest fear many publishers have is losing control. They do not want automation to introduce the wrong deals, creatives, or commercial terms into their workflow. AdCP sales agents are not meant to replace that judgment. Instead, they automate the coordination effort around it. Every deal comes down to a proposal that the publisher team can review, question, and either approve or reject based on inventory strategy, quality standards, and business fit.
AdCP sales agents can remove much of the operational burden from the workflow. But teams still need to step in manually to ensure everything runs as intended. In practice, that usually falls into three areas:
Configuration. Publishers still need to prepare the inventory and commercial information that the agent works with, define workflow rules, and keep that setup up to date.
Decision-making. Teams still need to review the deals that require human judgment, approve creatives, and step in when requests fall outside standard rules.
Monitoring and support. Publishers still need to watch how campaigns and workflows are progressing and support follow-up tasks such as billing and exception handling.
In practice, sales teams review recommendations and approve key decisions. The agent supports the coordination around those decisions by structuring requests, routing information, and keeping the workflow moving.
Real workflow testing shows where AdCP fits best
Practical testing makes the most sense for publishers that already run direct deals and want to understand whether parts of that workflow can be standardized without losing control. It is especially relevant when teams are spending too much time on repeated coordination across requests, approvals, and deal handling.
AdCP sales agents are not a new source of guaranteed demand, nor do they replace direct relationships, negotiation, or publisher judgment. Their value lies in helping publishers automate coordination tasks within existing direct-demand workflows, reduce manual overhead, and create a more structured way to process deals.
Early testing does not require a large implementation effort. What matters more is having direct deal experience and a clear interest in understanding how workflow automation can support the business.
Publishers can also engage with the AdCP community to better understand how agent-driven workflows are evolving and where this model can help automate existing direct-demand relationships. It is also a practical way to test how AI can support inventory monetization, reduce manual direct-sales work, and learn how AdCP is being adopted across the market.
