TLDR IT 2026-05-28

For years, IT tools forced teams into reactive, fragmented work. Console flips that, operating in the agent layer across your existing stack with end-to-end automation, AI-native ITSM, and natural-language deployment.

[This Live demo] walks through exactly how it works in practice

Register to see:

✅ Workflows built end-to-end in natural language by anyone

✅ Tickets triaged and follow-ups handled before a human steps in, with similar requests auto-resolved

✅ Apps connected at the agent layer to provision access and execute actions, governed by your rules

[Save your spot →]

Dell secured a $9.7 billion Pentagon agreement that will centralize procurement of Microsoft software, cloud subscriptions, and enterprise licensing across the Department of Defense, intelligence community, and Coast Guard. The deal highlights how large-scale government IT modernization is increasingly revolving around consolidated cloud, identity, and productivity platforms rather than standalone hardware refresh cycles.

The FBI warned that members of the Silent Ransom Group are now physically visiting law firms while impersonating IT staff, convincing employees to let them plug USB drives into corporate devices after phishing attempts fail. Modern social engineering is increasingly blending digital and physical tactics, forcing IT teams to rethink office access controls, USB policies, and how employees verify internal support staff.

Companies are pouring money into AI training programs, but many are overlooking the harder problem: whether employees actually have the curiosity, adaptability, and workflow context needed to use AI effectively day to day. Broad “everyone learns AI” initiatives often fail because organizations treat AI adoption like compliance training instead of redesigning workflows, incentives, and operating models around how work actually gets done.

The widely reported “AI pullback” is partly a category error: AI demand is split across supply-constrained infrastructure capex, shifting enterprise spending, canceled pilots, and rapidly growing metered token usage in production. Pilots are being killed while production deployments and token consumption continue to rise. The real constraint is measurement rather than demand.

DataGrail analyzed 2,400 software vendors and found that 63.6% of companies advertising AI capabilities did not disclose all third-party AI subprocessors in their legal documentation. The report highlights a growing operational problem for IT and privacy teams: vendors quietly routing company and customer data through multiple AI models and agent workflows that were never reviewed through procurement, security, or compliance processes.

What do Intuit, LinkedIn, and Intel have in common? They all use Scribe — the workflow AI platform that makes it easy to document, standardize, and scale the best processes. Fewer help desk tickets, faster onboarding, 93% less time spent on documentation. See how enterprise IT teams use Scribe, or start with a personalized demo.

Salesforce's Data 360 MCP Server, available in Developer Preview, lets any MCP client connect directly to Data 360. It provides a single interface for AI agents to access trusted context (setup, identity resolution, segmentation logic, transforms, and calculated insights) and speed up enterprise data onboarding. General availability is scheduled for this year.

Precisely launched Google Sheets support for Automate Studio, allowing business teams to run governed SAP workflows directly from spreadsheets while preserving enterprise validation and audit controls. The update reflects the growing push to bring enterprise automation into familiar collaboration tools instead of forcing users into specialized ERP interfaces.

Salesforce outlined several common implementation failures it sees across enterprise deployments, including weak data architecture, over-customization, poor integration planning, and unclear ownership between business and IT teams. The piece is a useful reminder that CRM projects often fail less because of technology limitations and more because organizations underestimate the operational governance and process design needed to run Salesforce as a long-term enterprise platform.

The piece breaks down common employee AI usage patterns that unintentionally expose enterprise data, including pasting contracts, source code, customer information, meeting notes, and internal documents into public AI tools. Rather than arguing for blanket AI bans, it pushes for a more operational approach centered on prompt-level controls, AI governance, approved tooling, and visibility into how employees are actually using generative AI across the business.

The article explains why “agentic lakehouse” systems fail when AI agents query data at scale, and presents an architecture map designed to make enterprise data discoverable, queryable, and reasoning-friendly for agents. It defines four technical layers and the key decisions at each layer (storage, Iceberg table format, Polaris catalog/credentialing, and a semantic/optimization layer), showing how weaknesses propagate upward and impact reliability, security, and performance.

Microsoft previewed automatic device isolation in Defender for Endpoint, which allows compromised systems to be quarantined faster without waiting for manual SOC intervention during active attacks.

Microsoft still struggles in mid-market identity deployments.

Google is moving the Ask Gemini entry point directly into the main Meet interface, signaling how quickly AI meeting assistance is becoming a default part of enterprise collaboration workflows rather than an optional add-on.