Enterprises Need 4 More Layers to Scale 150,000 AI Agents Beyond MCP and A2A
Updated
Updated · InfoWorld · Nov 11
Enterprises Need 4 More Layers to Scale 150,000 AI Agents Beyond MCP and A2A
3 articles · Updated · InfoWorld · Nov 11
Summary
Fortune 500 companies are projected to run more than 150,000 AI agents within two years, but protocols alone will not get them there; enterprises need four added layers to move from pilots to production.
Collibra’s 2026 report says 76% of organizations hit critical roadblocks in the past 12 months because MCP mainly connects a single agent to tools and data, while A2A coordinates work among separate agents.
The missing layers are identity and authorization, a governed business context layer, business-process redesign, and agenticops observability to control access, preserve trusted context, and audit agent actions.
That matters because MCP and A2A standardize connections, not trust or governance; experts warn siloed deployments can accelerate bad outputs, security gaps, and compliance risks.
The broader shift is from task-specific agents to multi-agent business workflows, pushing enterprises to rebuild legacy processes and operating controls around AI-native collaboration.