TLDR IT 2026-05-18

Microsoft leads enterprise agent orchestration with 38.6% adoption, followed by OpenAI at 25.7%. Anthropic has captured a 5.7% foothold. Enterprises are prioritizing security and governance over model choice, shifting focus toward the control plane to manage agent workflows, permissions, and auditability while avoiding vendor lock-in.

MDASH is an AI agentic system that autonomously scans for and fixes software vulnerabilities. It achieved 96–100% recall on real-world Windows bugs that previously required Patch Tuesday patches. This is a shift from AI-assisted to AI-autonomous cyber defense.

Anthropic's "skills" are driving B2B AI adoption, signaling a shift from consumer app stores to internal agent workflows. As SaaS companies prioritize integration, "we have a skill for that" becomes essential. Meanwhile, public SaaS valuations show a 3.1x median revenue multiple, with high-growth firms reaching 14.7x.

A volatile geopolitical climate and fractured regulatory landscape are forcing CIOs to shift from centralized IT strategies to highly adaptable, regional architectures. Global tech stacks are being redesigned with zero trust parameters, hybrid-cloud localized hosting, and "AI Councils" to navigate conflicting data compliance laws. To survive supply chain shocks and budget pressures, IT leaders must eliminate vendor concentration risk and treat data sovereignty as a core engineering requirement rather than a legal afterthought.

Docker has introduced Custom MCP Catalogs and Profiles to standardize enterprise AI tooling. Catalogs enable organizations to distribute trusted, curated server collections as OCI artifacts. Profiles allow developers to create, persist, and share portable, context-specific server configurations, improving workflow efficiency and reducing context window bloat for AI agents.

Amazon Q now supports document-level access control lists (ACLs) for S3 knowledge bases, enabling granular security for sensitive files. Administrators can use global ACL files or per-document metadata to enforce ALLOW and DENY policies, ensuring users only access authorized content while maintaining strict compliance and data governance.

Large organizations often manage over 1,000 APIs, leading to significant sprawl and duplication. By mapping technical operations to business capabilities, teams improve reusability and AI-readiness. Implementing these abstractions via wrappers or declarative workflows helps align engineering outputs with business outcomes, despite challenges in ownership and granularity.

Frontier labs are pouring billions into armies of forward-deployed engineers to manually force AI into messy corporate workflows, but this creates a costly consulting model rather than a scalable product. Human engineers are only a temporary bridge to discover a company's unwritten operational quirks during the initial pilot phase. True enterprise ROI only happens when this discovery is hardcoded into "forward-deployed software" that independently memorizes corporate context and handles exceptions over time.

Agent hooks provide deterministic control by attaching user-defined handlers to specific lifecycle events like PreToolUse and Stop.

Cybercriminals are increasingly using AI to automate phishing, malware creation, and even zero-day vulnerability exploitation at unprecedented speed.