TLDR IT 2026-10-01
Always-On AI 🤖, DIY SaaS 🛠️, Security Rethinks Identity 🛡️
Introducing Dots (7 minute read)
OpenAI has introduced "Dots," an always-on AI agent platform powered by GPT-6 Astra that operates its own cloud computer and browser to assist users across various tasks. Dots can connect to over 4,000 apps through plugins, learn user preferences over time, and execute background tasks such as bug fixing and content drafting. Dots are beginning to roll out across Pro, Business Premium, and Enterprise plans, with specialist dots also previewed for organizational use.
SaaS companies' newest rivals are their own customers (6 minute read)
Companies including Spotify, Twilio, and West Monroe are using AI to build internal tools that previously would have meant buying another SaaS product. It's an important shift for IT leaders: build-vs-buy is getting recalculated as custom internal software becomes dramatically cheaper and faster to create.
The problem with wild code? IT can't see it coming (5 minute read)
AI-assisted software development is making it easier for employees to build applications and automations outside traditional development processes. The bigger enterprise challenge may be visibility: knowing what has been built, what data it touches, who owns it, and whether security and lifecycle controls exist.
AI's Third Wave: Coworkers Break the Security Model That Worked for Agents (4 minute read)
Persistent AI coworkers can operate continuously with standing access rather than receiving temporary permissions for one task. That changes the identity problem for IT and security teams, making lifecycle management, least privilege, visibility, and offboarding increasingly important for non-human identities.
The Case for Meta Enterprise Platform (8 minute read)
Meta's newly announced Meta Enterprise Platform serves a crucial strategic purpose by acting as a secondary demand sink for the company's massive artificial intelligence capital expenditures. By offering its AI technology stack and agents to businesses, Meta can mitigate the financial risks of overbuilding compute infrastructure for consumer products while countering the market dominance of leading AI labs.
Whatever happened to the 36-month IT security roadmap? (6 minute read)
Fixed multi-year security roadmaps are becoming harder to maintain, as threats, business priorities, and technology change faster. Some security leaders are instead moving toward tiered priorities and continuously revisiting the roadmap while preserving longer-term strategic direction.
Stop Chasing AI Speed, Fix the Rework First (9 minute read)
Measuring AI success primarily through speed or hours saved can hide rework and poor outcomes. Focus on business results and explicitly define where AI assists, where human judgment remains necessary, and how employees participate in redesigning workflows.
How decades of disconnected modernization created OT security debt manufacturers can't escape (13 minute read)
Decades of disconnected modernization have created a severe OT security debt crisis for manufacturers, as new technologies were built on legacy systems without proper network segmentation or security architecture. Flat networks, default administrator accounts, and permanent remote access links have vastly expanded the attack surface, turning single footholds into enterprise-wide risks. Addressing this exposure requires asset visibility, realistic exposure management, and tailored controls for brownfield environments rather than complete system replacements.
Announcing FalkorDB 6.0 (9 minute read)
FalkorDB announced the release of version 6.0, the first version built on its new Rust engine. The company states that the new engine executes 39% fewer CPU instructions across benchmark queries, resolves 226 bugs, and introduces a new replication format called effects v3 while maintaining backward compatibility with existing Cypher queries and integrations.
Introducing skills in the Gemini app and Workspace, plus what's next for Gems (5 minute read)
Google is introducing reusable Gemini "skills" that can be combined inside prompts and shared across workflows, with Workspace rollout beginning October 5. Skills will eventually replace Gems, so Workspace admins should start planning for the transition and reviewing existing Gemini and Workspace Studio use cases.
MongoDB launches Atlas Agent Engine, Atlas Infinite as MongoDB 9.0 goes GA (3 minute read)
MongoDB has announced the public preview of Atlas Agent Engine, a governed agentic platform featuring sophisticated memory architecture, and Atlas Infinite, a deployment option that separates compute and storage to improve scaling. Additionally, the company is making MongoDB 9.0 generally available, bringing higher transaction throughput and faster updates. Together, these products aim to accelerate how enterprises move AI agents into production.
80% of network pros are OK with giving AI an autonomous role in network operations (5 minute read)
A Cisco survey of 1,000 IT professionals found significant willingness to let AI perform autonomous network operations. The trend suggests network management is moving beyond AI-assisted troubleshooting toward systems that can potentially diagnose and remediate issues with less direct human intervention.
NetApp ticks another hyperscaler box with Oracle Cloud Infrastructure deal (4 minute read)
Oracle and NetApp are planning a fully managed, OCI-native storage service built around NetApp ONTAP. The offering is intended to let enterprises move databases, VMware, regulated workloads, and AI data into OCI while retaining familiar NetApp management and protection workflows.
Anthropic revelations suggest a much stronger AI negotiating stance for enterprise CIOs (5 minute read)
Analyses of a confidential prospectus from Anthropic suggest that enterprise CIOs may hold a stronger negotiating position than previously assumed due to revenue concentration and a lack of long-term customer contracts. Vendors like Anthropic depend heavily on enterprise revenue, giving buyers leverage to negotiate price protections, data portability, and advance notice for model retirements. However, enterprises must also manage infrastructure dependencies and cost structures as frontier AI providers capture a dominant share of market revenue.
Get to know OpenSearch 3.9 (15 minute read)
OpenSearch 3.9 introduces improvements across search, observability, and cluster resiliency, highlighted by a faster native C++ engine for neural sparse vector search and expanded vector storage options like 16-bit half_float types and 2-bit through 4-bit scalar quantization. The release also brings a unified alerts view out of preview, adding anomaly detection and forecasting to help users triage signals across data sources from a single place.
Curated news 🗞️ and trends 📈 in IT strategy 💻, information security 🔐, and cloud computing ☁️.
Join 570,000 readers for
one daily email