TLDR IT 2026-09-16

AI is accelerating how software gets built and shipped. But for enterprise teams, moving faster also means rethinking security, governance, reliability, and control.

At {unscripted} Virtual, hosted by Harness, leaders from Anthropic, Workday, SpaceXAI, Riot Games, Google, and more will share how organizations are adapting software delivery for the agentic era, including how to:

✅ Scale AI across the SDLC with the right guardrails

✅ Keep security and governance aligned with AI-driven velocity

✅ Build platforms ready for increasingly autonomous agents

September 30 | Virtual | 10:00 AM PT

View the agenda →

Register →

Slack is expanding far beyond chat with Slackbot in channels, full-screen AI workspaces, live Salesforce-powered Surfaces, and Slack Code, where developers and AI agents can build and review code collaboratively. Slack is becoming a control surface for more business systems and agents, raising new questions around app governance, permissions, licensing, and what employees should be allowed to connect.

Microsoft has released an out-of-band Windows 11 update after September's Patch Tuesday caused problems with Remote Desktop, audio, virtualized apps, and other components. KB5129195 fixes several of the issues, but reported Explorer crashes, File History failures, some USB audio problems, and AMD GPU instability remain worth watching before accelerating enterprise deployment.

Despite dramatic improvements in developer productivity and falling unit costs for AI intelligence, software is becoming more expensive due to rising software scope and Jevons paradox driving up consumption. Software has historically been priced on the value delivered to the buyer rather than production labor costs. Consequently, efficiency gains are absorbed by increased inference usage and higher feature expectations rather than lower bills.

Salesforce's Dreamforce announcements point toward enterprise software being accessed increasingly through agents, MCP, and conversational interfaces rather than traditional application screens. The unresolved IT question is economic: as software moves away from predictable seat-based licensing toward agent usage, MCP calls, and consumption pricing, CIOs may gain flexibility while losing some cost predictability.

AI is turning business users into application and agent builders faster than traditional IT processes can keep up. Instead of trying to centrally build everything, CIOs increasingly need to own the foundations - identity, data architecture, observability, integration standards, and agent governance - while letting business teams build safely on top of them.

Building an inference stack takes months, especially in today's severely supply constrained environment. AMD Instinct™ Coder, powered by Spectro Cloud, ships as a Supermicro server with AMD Instinct GPUs and PaletteAI Inference Launchpad already flashed. Serve inference in a day.

Zoom is expanding its platform with Common Room integration and new Revenue Accelerator capabilities for engagement and forecasting. It is another example of collaboration vendors broadening into adjacent enterprise workflows, potentially giving IT teams more opportunities to consolidate tools, but also more overlap to evaluate across CRM, meeting intelligence, and sales platforms.

Salesforce Builder Central is a no-code AI workspace designed to let admins and business users build and configure apps and agents using plain language. Powered by Headless Toolkit, the workspace natively reads company metadata and Data 360 context while providing built-in trust, security, and automatic safety checks. The beta version is scheduled to launch the week of September 21.

HubSpot is adding centralized controls for which apps and AI connectors can be installed, who can install them, what data they can access, and who owns each integration. Particularly useful for IT: admins can require approval, use more granular OAuth scopes, restrict MCP access to company-provisioned accounts, see app activity logs, and get warned when deactivating someone who owns an integration.

Westpac's Adapt is a new enterprise data platform built on Microsoft Azure that consolidates data from 285 source systems into a single governed environment. The platform required migrating over 1PB of data and 14,000 data pipelines. Additionally, Westpac utilizes a Snowflake-powered intelligence layer to run advanced AI models on data at rest without moving it.

Scaling knowledge management from a personal Obsidian graph to an enterprise-grade "company brain" requires AI agents to manage knowledge without corrupting it through hallucinations or conflicting updates. Versioned branches, object storage, and a typed graph schema can keep changes controlled, with human review governing merges.

AI workloads are disrupting traditional cloud commitments as GPU availability, data movement, egress fees, and unpredictable inference demand make long-term AWS, Azure, and GCP agreements harder to forecast. CIOs are increasingly separating AI infrastructure economics from general cloud spend, negotiating more flexible capacity terms, and deciding which high-cost workloads need to remain portable across providers

Researchers at Wiz report active exploitation of three JFrog Artifactory vulnerabilities, including an authentication-bypass bug and two flaws being chained together for privilege escalation.

Teams organizers can give trusted B2B and guest presenters production controls for events, reducing the need to provision external event-production partners as internal users.

Anthropic walks admins through new Claude Enterprise controls for model defaults, per-user spend visibility, usage reporting, and managing consumption-based AI costs without a dedicated FinOps operation.