What Does A Beta Test Of A Sell-Side Agent Look Like?

Why It Matters

By automating low‑value, customized transactions, sell‑side agents can unlock untapped revenue and improve efficiency across the programmatic ecosystem, reshaping how publishers sell inventory.

Key Takeaways

• AI agents automate small, segmented ad deals
• Weather Co. tests sell‑side agent with premium inventory
• Optable integrates Claude for data‑driven sales proposals
• Agents could bridge direct and programmatic channels
• Early pilots aim to reduce revenue leakage to intermediaries

Pulse Analysis

The ad‑tech landscape has long relied on machine‑learning models to fine‑tune floor prices and match ads to inventory, but the next evolution is agentic artificial intelligence that can act as a sales representative. Leveraging large language models, these sell‑side agents interpret buyer requests, assemble proposals, and negotiate terms without human intervention. The open‑source code donated by AdCP to the Prebid project provides a common framework, allowing publishers to experiment without building the stack from scratch. This democratization accelerates the move from passive optimization to proactive, conversational selling.

The Weather Company’s first agent focused on a curated set of high‑viewability formats and premium audiences, exposing them through authenticated access rather than the open‑auction pool. Human operators still validate pricing, but the agent streamlines discovery and proposal generation, turning otherwise buried inventory into searchable assets. Optable took a different route, embedding Anthropic’s Claude into its data platform to translate RFPs into audience recommendations, package data‑media bundles, and even trigger programmatic deal creation in real time. Both pilots demonstrate that AI can handle niche, low‑budget deals that traditional sales teams often ignore.

Industry observers see sell‑side agents as a potential ‘third channel’ that sits between direct and programmatic buying, capable of surfacing inventory within Google Ad Manager or across supply‑side platforms while preserving human oversight for final approvals. By reducing transaction costs and enabling natural‑language negotiations, agents promise to capture revenue that currently leaks to intermediaries and to accelerate time‑to‑market for customized campaigns. However, widespread adoption will depend on robust integration, transparency of AI decisions, and clear governance to mitigate bias—challenges that early pilots are beginning to address.