Nexxen Adopts Claude and Google AI Buying Protocols

Nexxen, the supply- and demand-side platform formed from the merger of Tremor and Unruly, is wiring its programmatic stack into the emerging class of AI agent protocols backed by Anthropic and Google. The move positions Nexxen's platform to be queried and transacted against by autonomous agents, as agencies push to automate the manual, labor-intensive layers of media buying.

The integration centers on standardized interfaces that let large language models and agentic systems interact with ad-tech platforms programmatically. On the Anthropic side, that means support for Model Context Protocol (MCP) — the open standard that lets AI models like Claude connect to external tools, data sources, and APIs in a structured way. On the Google side, it aligns with the company's agent-to-agent (A2A) framework and its broader push toward agentic interfaces for advertising workflows.

Why Agent Protocols Matter for Programmatic

Programmatic media buying is already automated at the auction level — bids clear in milliseconds via OpenRTB. But the orchestration layer above the auction remains heavily manual: campaign setup, audience definition, inventory discovery, budget pacing, optimization, and reporting still consume significant trader and ad-ops hours. Agentic protocols target that orchestration layer.

By exposing platform capabilities through MCP and similar interfaces, Nexxen effectively makes its DSP and SSP functions callable by an AI agent. In practice, an agency planner could describe a campaign in natural language — target CTV households, a budget, flight dates, performance goals — and have an agent translate that intent into platform actions: querying available inventory, configuring line items, setting frequency caps, and adjusting bids based on returned performance signals.

The technical significance is that these protocols standardize how agents discover and invoke platform functions. Rather than building bespoke integrations for each agency tool, a platform that supports MCP can be addressed by any compliant agent. That lowers the integration cost of connecting buy-side AI systems to sell-side and demand-side infrastructure.

The Buy-Side Automation Push

Agencies and holding companies have been vocal about deploying AI to compress planning and activation timelines. Several have built or licensed agentic systems intended to sit across multiple ad platforms. For those systems to actually transact, the platforms on the other end need machine-addressable interfaces — exactly what Nexxen is providing.

For a TIER 2 platform competing against larger DSPs and the walled gardens, supporting these protocols early is a differentiation play. If agency agents can reach Nexxen's inventory and demand as easily as any other endpoint, the platform stays in consideration as buying workflows shift from human-driven UIs to agent-driven API calls.

Implications for Video and CTV

Nexxen's strength skews toward video and connected TV, where its supply relationships and data assets are concentrated. CTV buying is particularly fragmented — inventory spans publishers, FAST channels, and programmatic exchanges, each with its own setup quirks. An agentic interface that can navigate that fragmentation programmatically is potentially more valuable in CTV than in commoditized display.

There are open questions the protocol layer doesn't yet resolve. Agentic buying still has to respect the same plumbing as human buying: VAST creative handling, ad pod construction, frequency management, supply-path transparency via sellers.json and the supply chain object, and verification for fraud and viewability. An agent that books a CTV pod still needs valid VAST tags and SSAI-compatible delivery downstream. The protocols govern how intent reaches the platform, not how the impression is ultimately served.

There are also governance concerns. Letting autonomous agents allocate budget at scale raises the stakes on guardrails — spend limits, brand safety constraints, and human approval checkpoints. Standardized protocols make it easier to plug agents in, but they also make it easier to plug in agents that behave unpredictably. Expect platforms to layer permissioning and audit logging around these interfaces.

The Bigger Pattern

Nexxen is not alone — MCP adoption has spread quickly across software categories, and ad-tech is now following. The strategic signal here is that the industry's major AI vendors, Anthropic and Google, are becoming de facto standards bodies for how agents talk to advertising systems. For ad-ops and engineering teams, the near-term takeaway is that platform integration roadmaps will increasingly include agent-protocol support alongside traditional APIs and the IAB/OpenRTB stack. Whether agentic buying delivers on its efficiency promise will depend on how well these interfaces map onto the messy realities of inventory, creative, and measurement.

Stay on top of video ad serving and programmatic. Follow Adelerate.