InfoQ covers a presentation from LinkedIn engineering detailing how the company platform teams are enabling multi-agent AI tools across the organization. The talk focused on the Model Context Protocol and how LinkedIn built infrastructure to support dozens of AI agents operating across different business domains.
The presentation outlined LinkedIn approach to agent governance, including access control, monitoring, and cost management for multi-agent workflows. Platform teams serve as the enabling layer, providing shared tool registries, evaluation frameworks, and deployment pipelines that individual product teams can use to build domain-specific agents.
LinkedIn use of multi-agent systems spans recruiting, content moderation, sales intelligence, and internal developer tools. The company reported that the platform-team model has accelerated agent adoption while maintaining centralized oversight of safety and cost.