ReleaseBytes
databricks Databricks Blog ·

Databricks Accelerates Agentic Media Buying with New Reference Implementation

blogaidatabricksmedia
feature announcement

Databricks has released a reference implementation and self-deploy accelerator for agentic media buying, enabling autonomous buyer and seller agents to transact business on its platform. This solution addresses the critical need for governed data, transactional state, identity, hosted models, and end-to-end observability, which are often bottlenecks in agent-based projects. The accelerator aims to provide a working blueprint for teams to deploy in their own workspaces, streamlining the media buying process which traditionally involves significant manual coordination. Buyer and seller organizations can adapt these agents to their environments and transact with partners over open standards.

  • →Agentic Media Buying Reference Implementation and Accelerator
  • →Hierarchical Agent Crew Architecture
  • →Streamlined Transactions for Buyers and Sellers
  • →Platform for Agent Operations
  • →Support for Industry Standards
Features (2) ›
  • Agentic Media Buying Reference Implementation and Accelerator

    Databricks offers a reference implementation and a self-deploy accelerator for agentic media buying, allowing autonomous buyer and seller agents to transact on the platform. This solution is built on Databricks and uses the IAB Tech Lab's open source SDK, offering a blueprint for organizations to deploy in their own environments.

  • Hierarchical Agent Crew Architecture

    Buyer agents are structured as 'crews' of specialist agents using CrewAI, with hierarchical levels for strategy delegation, channel specialization, and tactical execution like audience planning and deal execution. Databricks Foundation Model APIs and Unity AI Gateway facilitate model selection and management.

Enhancements (3) ›
  • Streamlined Transactions for Buyers and Sellers

    The solution addresses the friction in traditional media buying, which relies on manual coordination through emails and spreadsheets. By enabling AI agents to handle tasks like discovering inventory, negotiating price, and closing deals, it frees up human talent to focus on strategy and creative work.

  • Platform for Agent Operations

    Databricks provides a unified platform that manages essential operational components for agents, including transactional state, hosted foundation models, governed data access controls, identity and trust mechanisms, and end-to-end observability for debugging and auditing.

  • Support for Industry Standards

    The reference implementation supports IAB Tech Lab's Agentic Advertising Management Protocols (AAMP), including AdCOM for shared vocabulary and OpenDirect/OpenRTB deals for transaction protocols, ensuring interoperability between buyer and seller agents across the industry.

Read the original announcement →

https://www.databricks.com/blog/agentic-media-buying-cannot-scale-without-right-foundation-see-how-buyers-and-sellers-get

Related releases