The latest developments in agentic AI: autonomous agents, multi-agent systems, LangChain and LangGraph updates, agent frameworks, and production deployments across the enterprise.
Dynatrace forecasts FY 2027 ARR of $2.38B-$2.4B and reported Q4 FY26 ARR topped $2B on its earnings call. The company discussed expansion of agentic AI and logs, covered margins and risks, and addressed M&A involving DevCycle and Bindplane.
According to WWD, Tapestry, Inc., the parent company of Coach and Kate Spade, is using agentic AI to help customers narrow product choices while preserving in-store experiences. At a media briefing WWD attended, Mandeep Bhatia, senior vice president of Global Digital Product and Omnichannel Innovation at Tapestry, said, "That world is not here yet," referring to fully autonomous bots shopping on behalf of consumers. Tamara Pircz, vice president of e-commerce at Kate Spade, told WWD that agentic AI can infer signals from browsing behaviour to surface options, but it does not complete purchases for customers. WWD also reports a recent survey of 500 U.S. consumers finding only 5 percent would allow agentic AI to buy for them, and that AI-driven narrowing often leads customers to visit physical stores.
Top Down Ventures announced the final close of Founders Fund I at US$28 million, exceeding an original US$25 million target, according to a company press release distributed via CNW and reporting by The Globe and Mail and BetaKit. The Vancouver firm raised capital from more than 100 limited partners, the majority drawn from the managed service provider (MSP) ecosystem, per those reports. The fund held a first close in October 2024 and completed its final close in April 2026, the CNW release states. The CNW release states it has deployed capital into 12 portfolio companies and that its first exit, zofiQ, was acquired by ConnectWise, returning 5.3x to the fund, as reported by BetaKit and the CNW distribution.
Per a press release distributed via Business Wire and posted on Montreal Gazette, Boomi and Red Hat announced a strategic collaboration to deliver an integrated, enterprise-scale stack for deploying agentic AI. The announcement says the offering will combine Boomi's Agentstudio and orchestration components with Red Hat's AI runtime and observability services to address data sovereignty, governance, and AI cost control. The release highlights features including connecting agents to live enterprise data, deterministic guardrails via Boomi Agent Control Tower and Gateway, and a Kubernetes-native runtime from Red Hat AI for performance and cost optimisation. Editorial analysis: Companies combining integration middleware with Kubernetes-native runtimes commonly reduce deployment friction and give enterprises clearer levers for sovereignty, governance, and cost management when moving from prototypes to production.
According to Microsoft Security Blog, Microsoft announced a new multi-model agentic scanning harness codenamed MDASH that helped find 16 vulnerabilities across the Windows networking and authentication stack, including four critical remote code execution (RCE) flaws. Microsoft's post attributes the discoveries to an ensemble that orchestrates more than 100 specialized AI agents and reports strong benchmark results: 21 of 21 planted bugs found with zero false positives on a private test driver, 96% recall on five years of confirmed MSRC cases for clfs.sys and 100% recall in tcpip.sys, and an 88.45% score on the public CyberGym benchmark, which Microsoft says tops the leaderboard. Infosecurity reports that May Patch Tuesday fixed 120 CVEs overall, 16 of which Microsoft says were found by MDASH. CSO Online reports the system will enter a private enterprise preview in June.
NVIDIA announced the Blackwell Ultra platform and rack-scale GB300 NVL72 and HGX B300 NVL16 systems in a May 2026 company release, and NVIDIA marketing materials report up to 50x higher throughput per megawatt and 35x lower cost per token versus the prior Hopper generation (NVIDIA blog and press release). CryptoBriefing and other coverage note partner activity, including VAST Data integrations and cloud deployments by Microsoft, CoreWeave and Oracle Cloud Infrastructure. Editorial analysis: Agentic inference workloads emphasize long-lived context, memory bandwidth and low-latency data access rather than short-lived stateless throughput, shifting the bottlenecks that infrastructure teams must plan for.
According to n8n's press release and PR Newswire, SAP made a strategic investment that values n8n at $5.2 billion, more than double the startup's October 2025 valuation of $2.5 billion. The companies also signed a multi‑year commercial agreement under which n8n's workflow automation platform will be embedded natively inside Joule Studio on SAP Business AI Platform, the firms say. Per n8n and SAP statements, the integration will provide a visual automation canvas supporting no‑code, low‑code and pro‑code development, connect SAP systems to more than 1,000 integrations, and inherit SAP security and compliance controls. n8n says it has more than 1,400 enterprise customers and a community of 1.7 million monthly active builders. "This partnership brings n8n into one of the largest enterprise ecosystems in the world," said Jan Oberhauser, CEO of n8n, in the company blog.
NVIDIA announced the Vera CPU, a purpose-built Arm-based data center processor aimed at agentic AI workloads, and has already signed multiple early deployment partners. According to NVIDIA's press release and reporting by DatacenterD and Wccftech, early collaborators and buyers include CoreWeave, Meta, Oracle Cloud Infrastructure, Alibaba Cloud, ByteDance, Lambda, Nebius, and Nscale. NVIDIA's materials and DatacenterD report the chip contains 88 custom Olympus cores, uses LPDDR5X memory, delivers up to 1.2 Tbps of bandwidth per chip, and the company announced a 256-processor liquid-cooled Vera rack that NVIDIA states can sustain more than 22,500 concurrent CPU environments. NVIDIA's press release also claims Vera delivers twice the efficiency and 50% faster performance than traditional rack-scale CPUs. Multiple OEMs and system vendors are listed as manufacturing partners in NVIDIA's release and industry reporting.
AnySearch has launched a search infrastructure product purpose-built for AI agents and enterprise systems, the company announced in a PR Newswire release and on its website. Per the PR Newswire release, AnySearch exposes a single unified API that aggregates vertical, authenticated sources such as finance, legal, academic repositories, code hosts, and structured APIs, and the product is available across developer ecosystems including GitHub, skills.sh, ClawHub, SkillHub, and Glama with users receiving 1,000 free API calls per day. AnySearch's website publishes internal benchmark results showing 76.4% overall accuracy across three datasets (Frames, FreshQA, WebwalkerQA) using z-ai/glm-5.1 as the LLM, and reports an end-to-end agent latency of 47.8s, which it compares to competitor figures for Parallel and Brave. The AnySearch website also advertises privacy features including "no tracking, no telemetry, and no logging."
Seeking Alpha reports that DigitalOcean is experiencing a major AI-driven growth inflection, with management guiding for revenue growth to double by next year and raising full-year guidance to 27% revenue growth and 39% adjusted EBITDA margin. Editorial analysis: Industry observers note that rising inference demand from agentic AI typically increases demand for low-friction cloud capacity, a tailwind for providers serving developer and SMB workloads, but Seeking Alpha's author retains a "Neutral" rating citing valuation and potential long-term pressure from hyperscalers.
Road to VR reports that Meta updated its open-source Immersive Web SDK (IWSDK) to include an AI-driven "agentic workflow" for building WebXR experiences, integrating coding assistants such as Claude Code, Cursor, GitHub Copilot, and Codex. Road to VR quotes Meta: "In practice, agentic workflows mean the AI does more than generate code; it also tests and validates it." Road to VR also reports that Meta demonstrated the system by rebuilding a 2022 VR gardening demo in about 15 hours using IWSDK and existing assets, and that the project is available under an MIT license. Editorial analysis: This lowers technical friction for web-based VR by automating larger portions of development and validation, which could broaden the creator base for WebXR.
Reporting by IIoT World and Avanade materials shows AI agents embedded in shop-floor workflows delivered measurable results in pilot customers. Per IIoT World, one global snack food customer cut inventory by 20%, and an electronics manufacturer recovered $35 million in lost fees in a year. Accenture and Avanade, in collaboration with Microsoft, demonstrated an "agentic factory" system at Hannover Messe 2026 and said early validation is underway with Kruger and Nissha Metallizing Solutions, with planned general availability later in 2026 (Accenture newsroom; Yahoo/Tech reporting). Per IIoT World and Microsoft blog posts, demos included edge ML that adjusts line speed based on quality thresholds and agents that assist operators with diagnostics and guided troubleshooting. Editorial analysis: Industry adoption hinges on combining data, process work, and operator-facing interfaces.
CryptoBriefing reports that executives from Trust Wallet and Mesh presented a vision for AI-powered wallets at Consensus Miami 2026, arguing AI agents could automate trading, payments, and self-custody tasks. CryptoBriefing says Trust Wallet CEO Felix Fan framed wallets as potentially the "new browser" for crypto, and that Mesh CTO Arjun Mukherjee emphasized infrastructure to let agents move money programmatically, citing a Mesh update published on May 1. CryptoBriefing also flags security and privacy concerns, citing a CryptoRank report dated April 13 and guides published by Cobo on May 7 about vulnerabilities in third-party AI integrations. The Consensus agenda lists a session titled "The Agent Wallet Is the New Browser" for May 7, with Mukherjee as a speaker, per CoinDesk's event pages.
Cloudflare announced cuts of about 1,100 roles as it reorganizes around what executives called the "agentic AI era," according to a memo published on the company's blog and reported by the San Francisco Chronicle. The company reported first-quarter revenue of $639.8 million, up 34% year over year, per the Chronicle, but shares fell sharply in after-hours trading (more than 14%), according to Business Insider and Barron's. In the employee message, co-founders Matthew Prince and Michelle Zatlyn said the company has seen AI usage grow by more than 600% in the past three months, and that teams are running thousands of AI agent sessions daily, according to coverage by the Chronicle. The company estimated restructuring charges of $140 million to $150 million, with most charges expected in Q2 and the plan substantially complete by the end of Q3, per the Chronicle.
Business Insider reports that executives across tech and other industries are publicly citing metrics about how much code their companies generate via AI agents, including mentions of early adopters such as Anthropic, Meta, and Google. The article says the trend appears on earnings calls and in interviews, where leaders use AI-code volumes as a signal to investors and potential hires. Alex King, founder of AI talent firm ExpandIQ, is quoted telling Business Insider, "Visibly AI-forward companies attract the right talent profile needed to actually become an AI-centric company." Business Insider frames the shift as moving AI output metrics into executive communications and recruiting narratives.
SiliconANGLE reports from Appian World 2026 that enterprise deployments of agentic AI require process-level guardrails, governance, and human review to be useful in regulated environments. According to the coverage, Appian Corp.'s platform incorporates human review and approval so that AI agents cannot build or execute autonomously, a safeguard emphasised by Appian co-founder and CEO Matt Calkins during an interview on theCUBE. Calkins said, "Agents are an incredibly powerful tool, but they have to be used correctly... Agents actually need process more than any other form of AI." Reporting by SiliconANGLE also highlights a process-centric architectural approach that embeds agents into existing workflows to preserve auditability and scale AI across the enterprise.
IIR reports that Mitch Wein, Head of Insurance Strategy and Advisory at Datos Insights, says agentic AI is beginning to challenge longstanding assumptions about where underwriting, policy administration and other core-system functions should reside. According to IIR, Wein described the emergence of the MCP standard as a potential connective layer that can link AI agents to systems and data sources, enabling multi-step workflows that span models, platforms and operational domains while preserving context, access controls, lineage and governance. IIR quotes Wein saying, "The core system vendors are probably in trouble," and quoting him on earlier limitations: "You couldn't string together the workflows." Reporting by IIR also notes that Anthropic introduced MCP in November 2024 and donated it to the Agentic AI Foundation under the Linux Foundation in December 2025.
Cryptobriefing reports that AMD and Intel shares climbed to new all-time highs Friday as investor interest in AI infrastructure and rising CPU demand lifted chip stocks. Per Cryptobriefing, AMD shares rose to roughly $443, taking its market capitalization to about $720 billion, while Intel shares climbed to around $124, pushing its market value near $627 billion. Cryptobriefing attributes the rally to AMD's stronger-than-expected first quarter, with revenue of $10.3 billion (up 38% year over year) and data center revenue of $5.8 billion (up 57%), and to Intel's reported first-quarter revenue of $13.6 billion (up 7%) and non-GAAP EPS of $0.29. Cryptobriefing also quotes AMD CEO Lisa Su and Intel CEO Lip Bu Tan on agentic-AI-driven CPU demand, and notes reports of Apple talks with Intel and Samsung and expanded Intel-Google collaboration.
Hugging Face released an agentic toolkit for Reachy Mini, its open-source desktop robot, that generates, tests, and ships robotics code from plain-English prompts. According to the Hugging Face blog, the agentic flow lets users describe desired behavior and the agent writes, tests, and deploys the app to the robot without an SDK. The blog reports an install base approaching 10,000 units after "almost 3,000" units shipped last week and says another 1,000 units will ship in the next 30 days. The Reachy Mini App Store already hosts 200 apps from 150+ creators, per Hugging Face. Clément Delangue, co-founder and CEO of Hugging Face, is quoted in coverage saying, "For 60 years, robots were built by roboticists. As of today, they can be built by anyone," (reported by The Robot Report). Reachy Mini specifications, pricing (from $299), and kit variants are listed in product coverage.
Reporting from Reuters and The Information, aggregated by multiple outlets, says Meta is building an "agentic" AI assistant that can perform multi-step tasks for users. Reuters, citing the Financial Times, reports the assistant is powered by a new model called Muse Spark and is being tested internally. The Information reports an internal project codenamed "Hatch" that has been trialed on simulated third-party services such as DoorDash and Reddit. Multiple outlets, including Engadget and Digital Trends, report Meta aims to add a shopping-focused agent inside Instagram that can find and purchase items users see in Reels, with a target launch window reported as before Q4 2026 by The Information (as cited by Reuters). Digital Trends and others note this follows Meta's failed attempt to acquire Manus AI, a deal reportedly blocked by Chinese regulators.
IBM and Yotta Data Services announced plans to deliver an agentic AI platform for Indian enterprises and government organisations, the companies said in public statements and press coverage on May 5 and May 7, 2026 (Business Today, CRN, company releases). Per multiple reports, the platform will host watsonx Orchestrate on Yotta's Shakti Cloud infrastructure and intends to bring IBM Sovereign Core to India via Yotta, aiming to address data residency, governance, security, and regulatory compliance requirements (CRN; Business Today; VARIndia). The offering is described as enabling deployment and management of AI agents across IT service management, HR, finance, procurement, and customer support. "AI innovation in India must be anchored in sovereignty, security, and performance," said Sunil Gupta, co-founder and CEO of Yotta Data Services, in coverage by Business Today.
SmashingApps defines "agentic AI" as AI systems that autonomously execute multi-step tasks by planning, acting, observing results, and adjusting until completion, rather than only returning answers. SmashingApps cites examples including Cursor Background Agents, OpenClaw, and Anthropic's Claude Agents as 2026 agentic products that perform end-to-end workflows. SmashingApps also outlines four core components of an agent: a natural-language goal, a reasoning engine (an LLM), external tools (search, code execution, APIs), and memory / context. Editorial analysis: For practitioners, the shift toward agentic systems elevates integration, state management, and safety engineering over pure model prompt engineering.
According to an article republished on ITSecurityNews.info (from DZone Security Zone), the rise of agentic AI is undermining core assumptions about identity, access, and accountability. The article reports that traditional IAM, PAM, and SSO tooling were built around the premise that users are humans and that audit trails, binary authorization decisions, and session-based models no longer map cleanly to autonomous agent workflows. Editorial analysis: For practitioners, this trend implies a need to rethink authentication, authorization, and auditing primitives to handle non-human actors, ephemeral credentials, chained actions, and richer attestation metadata.
According to a CloudNativeNow interview with Traefik Labs CEO Sudeep Goswami, enterprises are juggling three migrations: lifting workloads off legacy virtualization, modernizing Kubernetes deployments, and integrating agentic AI. The interview reports Goswami's view that AI-generated code is reaching production faster than prior waves, making the runtime, not just CI/CD pipelines, the key control point. The piece highlights runtime governance primitives such as policy enforcement that travels with workloads, identity-aware routing for agent-to-service calls, and observability that treats AI traffic as a first-class category. The article also reports Traefik integrations across SUSE Rancher, RKE2, and the SUSE AI Factory, and notes Goswami's view that architecture choices made in the next 12 to 18 months will shape platform resilience under prolonged AI-driven change (CloudNativeNow).
Showing 24 articles
