Plumbing the
information age

Chronosphere differentiates through its control-plane architecture that proactively manages telemetry data and prevents cost escalation. The platform already serves observability for two of the top five leading AI frontier labs. A critical design pattern centers on data quality: the platform pre-computes as much as possible ahead of time, recognizing telemetry data is notoriously messy. For practitioners evaluating agentic AIOps, the Temporal Knowledge Graph offers context-aware AI agents that reduce investigation friction by connecting schema-diverse data, agent decision tracing, and historical patterns. Integration with Palo Alto Networks' security operations center strategy (XSIAM) positions observability data as foundational for both reliability and threat detection automation, moving beyond traditional reactive monitoring toward autonomous operations.

ServiceNow's AIOps solution leverages ITOM Health and Observability applications to ingest events, logs, metrics, and traces from monitoring tools, learns and models application behavior in real time, and uses AI/ML techniques to correlate and deduplicate alerts with generative AI enrichment. For practitioners considering ITOM modernization, these workshops provide direct access to experts on event management, metric intelligence, health log analytics, and synthetic monitoring. The focus on zero-outage operations reflects the broader industry shift toward autonomous IT platforms that move beyond reactive monitoring toward predictive and self-healing operations with policy-driven automation.

Gartner predicts 40% of enterprise applications will include task-specific AI agents by 2026 (up from less than 5% in 2025), and as systems move from pilots to production, orchestration layer selection becomes critical infrastructure decision. Most widely adopted open-source frameworks are LangGraph (stateful, graph-based workflows), CrewAI (role-based collaboration), and Microsoft AutoGen. Commercial platforms like Vertex AI Agent Builder and AWS Bedrock Agents provide managed orchestration with cloud integrations. For network practitioners, the critical distinction is architectural: a framework decides how agents reason, hand off work, and recover from errors, but does not determine governance, access control, or production cost—those belong to the infrastructure layer above. The first step toward reliable multi-agent orchestration at scale is establishing a shared context layer with governed business glossary definitions, lineage mapping, ownership assignment, and instrumented logging of all agent interactions. This maturation means network operations teams can now select purpose-built orchestration stacks without building custom coordination from scratch.

CrowdStrike announced the appointment of AJ Shipley as Chief Product Officer on July 15, 2026, leading the product organization to accelerate innovation across the CrowdStrike Falcon platform. As organizations deploy AI, every agent, model, and agentic workflow becomes a new attack surface to defend and a new identity to govern, creating two demands: using AI to defend enterprise through agentic SOC, and securing the AI enterprises depend on. Shipley will extend CrowdStrike's lead in securing the agentic era. Shipley brings over two decades of experience, joining from Splunk Security where he served as CPO. Previously he led product management for Cisco's Threat Detection and Response portfolio spanning XDR, EDR, NDR, email security, and vulnerability management, and built Cisco XDR from the ground up while driving due diligence for Cisco's $28 billion acquisition of Splunk. While the announcement was made Wednesday, Shipley joined the security vendor in May. The hire reflects CrowdStrike's architectural bet that unified endpoint telemetry, real-time threat intelligence, and expert validation position it uniquely to capture enterprise AI security as the dominant cybersecurity tailwind. BTIG raised its price target to $237 concurrent with the announcement, citing strong field demand for AIDR, Next-Gen SIEM, and Exposure Management across enterprise customers.

Microsoft Windows received the most patches this month with 413, followed by Extended Security Updates (ESU) with 335 and Microsoft Office with 95. CVE-2026-56155 is an Important elevation of privilege vulnerability affecting Active Directory Federation Services (AD FS) with a CVSS score of 7.8. An insufficient granularity of access control flaw allows a low-privileged local attacker to elevate privileges with no user interaction and low attack complexity, potentially granting administrator privileges. CVE-2026-57092 is a Critical elevation of privilege vulnerability affecting Microsoft Windows VMSwitch with a CVSS score of 9.9, indicating potential for hypervisor escape scenarios. Roughly half an hour after Microsoft's Patch Tuesday disclosures, details emerged of an unpatched privilege escalation vulnerability affecting the Windows User Profile Service on all currently supported Windows versions. The publicly released PoC is deliberately stripped down, but the author claims the original version imposes no restrictions and is not limited to UsrClass.dat, potentially allowing any registry hive to be loaded. Successful exploitation could enable registry-based persistence, credential theft, or security product tampering. For SOC and vulnerability management practitioners, the 622-vulnerability volume and post-disclosure zero-day indicate heightened patch cycles and rapid attacker pivot. Prioritize VMSwitch and AD FS testing in disconnected labs before broad enterprise rollout.

BTIG's field research flagged a strong demand environment with AI acting as an extra tailwind for CrowdStrike's core endpoint security business. The firm noted positive feedback on newer products like Next-Gen SIEM, Exposure Management, and AI Detection and Response. CrowdStrike generated about $4.81B in revenue over the trailing period with a strong 75% gross margin. Q1 2026 showed $590.94M in operating cash flow and $470.74M in free cash flow. With over $4.55B in cash and a current ratio of 1.5, liquidity is solid, providing runway for continued investment in growth and AI-driven security tools. In early July 2026, CrowdStrike expanded its AI-driven cybersecurity push with surging demand for its new AI Detection and Response (AIDR) offering, record new annual recurring revenue, raised full-year guidance, a 4-for-1 stock split, and fresh identity security collaborations. This cluster of AI-focused product traction and ecosystem partnerships suggests CrowdStrike is extending its role beyond endpoint protection into broader identity and managed security services. For SRE and AIOps practitioners evaluating consolidation platforms, Benchmark's upgrade reflects analyst consensus on machine-speed response and agentic automation as genuine operational levers. The 75% gross margin and strong cash position support continued investment in Next-Gen SIEM and AIDR, areas where traditional point solutions struggle with alert volume and context correlation.

On July 15, 2026, 3M (NYSE: MMM) and Microsoft (Nasdaq: MSFT) announced a strategic partnership focused on AI data center infrastructure and enterprise transformation. The technical focus centers on optical interconnect design: by using an expanded beam optical interface instead of the direct contact required in traditional connectors, EBO technology is designed to make fiber connections faster to install, more tolerant of contamination and easier to maintain, helping Microsoft reduce the need for frequent cleaning and inspection while supporting reliable optical performance in dense, high-volume deployment environments. For practitioners, this matters because deployment speed directly impacts data center time-to-production—Azure's pilot results suggest the solution reduces fiber connection time by nearly 80%, a meaningful operational win for hyperscaler buildouts. 3M is scaling production of its EBO technology to meet accelerating demand from hyperscalers and data center operators, has advanced single-mode expanded beam optical technology for high-volume data center applications, and helped establish the EBO Multi-Source Agreement (MSA) to support standardization and broader industry adoption. The partnership signals growing focus on physical layer optimization as a constraint alongside compute and switching.

This research introduces MCP-enabled agents for autonomous management of IP-over-DWDMnetworks, combining agent architecture patterns with optical network operations. The work addresses a critical gap in network automation: current systems rely on imperative scripts or simple rule engines, while this approach deploys reasoning agents that can handle intent-based provisioning and autonomous problem resolution across the full network lifecycle. The paper validates the approach through simulation of real provisioning workflows, showing how agents can decompose complex network tasks (configuration validation, SLA optimization, fault recovery) into sequential steps with tool invocation through the MCP standard. For network operations teams, this signals a transition from monitoring-centric alerting to autonomous remediation systems that can act independently on infrastructure problems. The architectural pattern—using Claude or similar LLMs as the reasoning core with gNMI/NETCONF tools for network interaction—is production-relevant for operators managing large optical networks. Notably, the work stays within standard protocols (gNMI, NETCONF, Yang models) rather than requiring proprietary extensions, making adoption feasible for existing network stacks. Caveats: demonstration is simulation-based; real-world deployment would require extensive safety validation and fallback mechanisms to prevent cascading failures from agent mistakes.

Inkling is a 975B-parameter MoE system with 41B active parameters trained on 45 trillion tokens of text, images, audio, and video, with native reasoning across all four modalities. It supports up to 1M-token context windows. Unlike competitive open-weight models, Inkling emphasizes calibration and tunable reasoning effort rather than raw benchmark dominance; Thinking Machines explicitly states Inkling is not the strongest overall model available today, open or closed. The model debuts at 41 on the Artificial Analysis Intelligence Index, the leading open-weights release from a U.S. lab, scoring 3 points higher than NVIDIA's Nemotron 3 Ultra. On agentic tasks, Inkling scores 1238 Elo on GDPval-AA v2, higher than Kimi K2.6 (1190) and DeepSeek v4 Flash (1189). For operations teams, token efficiency is key: averaging 25K output tokens per Intelligence Index task compared to 43K, 38K and 37K by competitors. Thinking Machines used Moonshot AI's Kimi K2.5 to generate early post-training data, though commits to fully self-contained post-training next. Available now via Tinker (64K/256K context), Hugging Face open weights, and third-party inference providers.

Kimi K3 is a Mixture-of-Experts model with roughly 2.8 trillion total parameters and a 1-million-token context window, with two variants—K3 Max for chat and agent tasks, and K3 Swarm Max for large-scale parallel processing. The architecture introduces Kimi Delta Attention (KDA), a hybrid linear attention mechanism that enables up to 6.3x faster decoding in million-token contexts. Sparsity mechanisms include Stable LatentMoE, effectively activating 16 of 896 experts. On independent benchmarks, Artificial Analysis puts K3 at 1668 Elo on long-horizon knowledge work evaluation (+732 from Kimi K2.6), with 1547 Elo on agentic tasks. Moonshot's benchmarks show K3 mostly beating Claude Opus 4.8 max and GPT-5.5 high, while losing out to Claude Fable 5 and GPT-5.6 Sol. The API is available now; full model weights are promised by July 27. The competitive threat is stark: Moonshot is offering K3 at prices well below premium models, raising questions about how long U.S. labs can charge top dollar for frontier-level intelligence.

Enterprise AI adoption is entering a new phase as organizations move from pilots and experimentation toward measurable business impact, according to findings from Info-Tech Research Group. The research, based on 551 survey responses, found that 42% of organizations have achieved department-wide AI adoption with measurable impact, with enterprises having a dedicated, governed AI strategy three times as likely to report measurable AI impact compared to those with no active strategy. Measurable value is significantly more likely when organizations have a dedicated AI strategy, strong data readiness, clear executive ownership, and business cases focused on productivity, risk, quality, and revenue rather than cost reduction alone. This signals a fundamental maturation away from broad-based experimentation toward disciplined, outcome-focused deployment—a critical inflection point for practitioners evaluating vendor claims and internal ROI justifications.

AI models are becoming ever more capable, but exactly what enterprise adoption will look like remains a big question. In a bid to shape that future, labs like Anthropic and OpenAI have spun up separate businesses dedicated to deploying AI engineers to their customers' offices—a bet that assisting businesses in figuring out how to use their AI models is the next trillion-dollar category. Ode with Anthropic is the $1.5 billion, AI implementation company that the AI lab launched in May as part of a joint venture with Blackstone, Hellman & Friedman, Goldman Sachs, and others. According to Ode executives, model selection matters, but it's not where the majority of calories are spent—it's one ingredient in a system that has to be engineered. The founding belief behind Ode is that non-AI companies are going to be among the big winners of this whole AI moment if they adopt the technology the right way, but to take AI and rewire core business processes or customer experiences with it requires a lot of help and top-caliber applied AI talent, which is not something most companies have. This represents a structural shift in how frontier labs now view enterprise value capture—less model licensing, more applied delivery at scale.

AI agent startup funding in July 2026 reached $1.8 billion across 12+ deals, led by enterprise automation and developer tools categories. Sequoia Capital, Index Ventures, and Andreessen Horowitz dominated deal flow, while average valuations climbed 40% quarter-over-quarter to $280M. Enterprise automation agents captured 58% of total capital, reflecting investor preference for B2B monetization over consumer plays, while developer tooling agents raised $420M with an average Series A valuation of $185M, signaling strong product-market fit in coding assistance. Harvey AI led July 2026 with a $200M Series C at $2.1B valuation, followed by Lovable's $200M Series B at $2.8B, Glean's $180M Series D at $2.7B. In July 2026, 62% of deals are Series B+ rounds averaging $150M with established revenue traction ($25M+ ARR), reflecting the category's evolution from experimentation to proven business models with measurable ROI.