MIT Technology Review Insights and Microsoft surveyed 300 technical experts across 12 industries to produce the 2026 Agent Confidence Index, rating confidence in AI agents across 101 tasks. Average confidence sits at 64/100, with automated report generation (83.5), boilerplate code generation (82.5), and certificate expiration monitoring (81.5) leading. Even complex tasks like service mesh configuration and database schema migration are gaining traction. Key findings: 59% of respondents prioritize keeping humans in the loop, 80%+ see meaningful career opportunity in agent-augmented roles, and agents perform best in well-integrated environments with unified data and governance. Microsoft positions its Agent Platform, GitHub Copilot, Azure SRE Agent, and Microsoft IQ as the infrastructure for enterprise-scale agentic AI.
