A new report from Gartner and McKinsey confirms what the enterprise software industry has been watching in real time: autonomous AI agents have crossed $500 billion in combined enterprise value across vendors, and deployment rates among Fortune 500 companies have reached 100% -- every large enterprise is now running at least one agentic workflow in production.
What Changed in 12 Months
Twelve months ago, AI agents were a proof-of-concept. Today, they are operational infrastructure. The shift accelerated dramatically after OpenAI's Codex deployment through seven major consulting firms, followed by Anthropic's Claude Agents API reaching general availability and Microsoft's Copilot Agents becoming default in Office 365 enterprise plans.
Where Agents Are Actually Running
The most common deployments are in software development (code review, test generation, PR management), customer support (Tier 1 resolution without human escalation), financial operations (invoice processing, reconciliation, regulatory reporting), and HR workflows (screening, onboarding, compliance documentation). Less common but growing fast: autonomous legal review, AI-driven M&A due diligence, and supply chain optimization agents that negotiate with vendor APIs directly.
The Infrastructure Winners
The agentic wave has created three clear infrastructure winners: Anthropic's MCP protocol (now the de facto standard for agent-tool connectivity with over 100 production servers), Vercel's AI SDK (processing 2 billion agentic API calls per month), and LangChain's orchestration layer (used in 68% of enterprise agent deployments). The losers are traditional RPA vendors -- UiPath and Automation Anywhere both saw share price declines exceeding 40% in Q1 2026 as AI agents rendered their legacy automation tooling obsolete.