xAI & Grok

xAI has released Grok 4.20 Heavy, featuring 16 specialized agents that pursue a big research project as a team. Users with the SuperGrok Heavy tier now tap into this full power. Once asked, Grok 4.20 Heavy fires up multiple agents to tackle queries from every angle, with the standard version using a core team of four agents that deliver improved responses compared to single-model AIs and combine everything into a compound answer.

xAI launched Grok 4.20, and it packs in a team of agents that work together on every request. Users watch them reason step by step through a new live thinking interface. When you ask something, the agents kick in together. Grok leads the way while Harper, Benjamin, and Lucas each grab a piece of the job that matches their focus. They split the problem, think in parallel, and trade ideas back and forth.

At least nine engineers, including two co-founders, have announced their exits from xAI in the past week, fueling online speculation and raising questions about stability at Musk's AI company amid mounting controversy. At an all-hands meeting Tuesday night, Musk suggested the exits were about fit, not performance. "xAI was reorganized a few days ago to improve speed of execution," Musk wrote.

OpenAI launched Frontier, a new platform designed for enterprises to build and deploy agents while treating them like human employees. On Thursday, AI giant OpenAI announced the launch of OpenAI Frontier, an end-to-end platform designed for enterprises to build and manage AI agents. It's an open platform, which means users can manage agents built outside of OpenAI too.

More than half of organizations now deploy AI agents for multi-stage workflows, with 16% running cross-functional processes across multiple teams. The numbers tell a clear story: 80% of enterprises report their AI agent investments already deliver measurable economic returns. This is production-grade deployment showing real results.

OpenAI Agents

Y Combinator-backed Typewise launched its AI Supervisor Engine for multi-agent orchestration in customer service. The system coordinates teams of AI agents handling customer requests via email, chat, and messaging with human handoff capabilities.

NIST launched the AI Agent Standards Initiative to ensure secure, interoperable AI agents capable of autonomous actions. The initiative aims to foster industry-led standards while maintaining U.S. technological leadership in autonomous AI systems.

Snowflake and OpenAI announced a $200 million partnership to integrate OpenAI models directly into Snowflake's data platform. The collaboration enables organizations to build agents that analyze proprietary data and execute complex workflows with built-in governance.

Anthropic & Claude

Anthropic unveiled its new enterprise agents program on Tuesday, aiming to integrate agentic AI into everyday workplaces. The program includes pre-built agents for common enterprise tasks like financial research and engineering specifications through a plug-in system. The system builds on Claude Cowork and plug-ins announced in research preview on January 30th, focusing on easier deployment within companies with private software marketplaces and controlled data flows.

The Briefing livestream took place on February 24, 2026, featuring product and engineering leaders from Anthropic demonstrating what's possible when Claude knows your work environment. The event included deep dives into how Cowork and plugins work in enterprise deployments and showed how different Anthropic teams use these tools in their daily work with live demos.

Just weeks after AI tools shook software stocks, Anthropic is pushing deeper into workplace automation by updating Claude AI to perform better at specific job tasks in design, HR, and wealth management through Claude Cowork launched in January. The rapid-fire updates have given Wall Street whiplash, making investors nervous about AI making other businesses' products obsolete, though Anthropic denies trying to replace enterprise software products.

Claude Sonnet 4.6 is now available in Amazon Bedrock, offering frontier performance across coding, agents, and professional work at scale while approaching Opus 4.6 intelligence at lower cost for high-volume use cases. New open-source Agent Plugins for AWS were introduced, extending coding agents with skills for deploying applications to AWS and generating architecture recommendations directly from coding agents.

Microsoft Agent Framework is now in Release Candidate status for both .NET and Python, marking API surface stability and completion of all features intended for version 1.0. The comprehensive, open-source framework provides a unified programming model for building, orchestrating, and deploying AI agents with multi-provider support including Microsoft Foundry, Azure OpenAI, OpenAI, and Anthropic Claude.

Entro Security announced Agentic Intent Monitoring for Claude Code, providing security teams real-time visibility into AI agent sessions including user interactions, prompts, and MCP server requests to solve the audit trail gap enterprises face with agents. The announcement came at the second annual NH(A)I Global Summit in New York City on February 26, 2026, exploring emerging risks and governance controls for enterprise AI agents.

NIST's Center for AI Standards and Innovation announced the AI Agent Standards Initiative on February 17, 2026, to ensure AI agents capable of autonomous actions can be adopted with confidence and interoperate smoothly across the digital ecosystem. Working with federal partners, the initiative aims to foster industry-led AI standards while cementing U.S. technological dominance as AI agents can now work autonomously for hours and manage complex tasks.

Microsoft Copilot

Copilot chat is now available directly inside apps built with Power Apps, bringing the intelligence of Microsoft 365 Copilot into the flow of business processes. This unified experience—currently limited to model-driven apps—lets users ask questions, reason over in‑app data, and connect insights from documents, communications, and collaboration—without leaving the application they're working in. Teams can also leverage powerful first‑party agents like Researcher and Analyst, as well as custom Copilot Studio agents, to analyze data, generate insights, and take informed action directly within their apps.

Organizations can now understand which agents are in use, how they're performing, and what they cost across environments. Admin controls are designed to align agent behavior with intended use, while agent evaluations support ongoing quality and improvement. Paired with Microsoft Agent 365, organizations get a unified view of agents across Microsoft 365 Copilot and Copilot Studio, giving business and IT leaders the clarity needed to scale with confidence.

Microsoft is continuing to iterate quickly on its AI assistant stack with a fresh Copilot Chat roadmap for February 2026, focused on tightening the experience rather than just adding more features. The latest update delivers one‑click email summarization in Outlook and previews higher‑quality, faster image generation powered by the GPT‑Image‑1.5 model for Microsoft 365 users.

Susanth Sutheesh's recent YouTube update walks viewers through the February 2026 round of changes to Microsoft 365 Copilot, describing 36 feature additions, improvements, and policy updates across apps and enterprise controls. Copilot Agents and extensibility updates let teams automate repeatable workflows and integrate custom AI behaviors.

Based on customer feedback, we've now increased the character limit for the Copilot input from the previous 500 characters to 10K in all Copilot surfaces, including Standalone, Report pane, Apps, Mobile, and Embed. This will allow you to type or copy/paste longer and expressive prompts to Copilot to meet your needs. Fabric Copilot capacity enables organizations to consolidate Copilot usage across Power BI Desktop, Pro and Premium per-user workspaces into a single designated capacity.

We break down ten common agent misconfigurations we observe in the wild, showing how to detect them using Microsoft Defender Advanced Hunting via the relevant Community Hunting Queries, and mitigate them using safe configurations with Copilot Studio. Copilot Studio agents are restricted to a Power Platform environment, which enforces data, security, and governance policies. Properly managing environments helps prevent risks like oversharing and controls who can create or deploy agents.

We've launched the new agent architecture guidance hub and a refreshed Microsoft Copilot Studio guidance hub. These on-demand resources offer end‑to‑end documentation across the agent lifecycle—from design and planning through operations, governance, and advanced architectural patterns. Built on established practices from Microsoft engineering teams and real‑world deployments, these hubs give architects, developers, and IT a shared blueprint to work from.

Google & Gemini

Google Cloud released its 2026 AI Agent Trends Report revealing how agentic AI will reshape business. The report highlights multi-agent systems that collaborate on complex workflows, predicts agents will automate routine tasks freeing employees for strategic work, and showcases partnerships like Salesforce's Agent2Agent protocol for cross-platform AI collaboration.

Google launched new AI subscription tiers with enhanced agent capabilities. Google AI Ultra ($249.99/month) provides 20x higher limits for multi-agent workflows, while Jules coding agent gets enhanced capabilities. The plans include browser agent prototypes and expanded agent request limits for enterprise use.

Google published its 2026 Responsible AI Progress Report detailing governance frameworks for agentic AI systems. The report covers seven-layer governance spanning research, policies, testing, and monitoring for agents that can act autonomously. Google is collaborating with the UK AI Safety Institute on advanced agent evaluation.

Deutsche Bank partnered with Google Cloud to develop agentic AI for trading surveillance and compliance monitoring. The AI agents will monitor trading activities, detect anomalies, and flag suspicious communications to human compliance officers. Multiple banks are testing similar agentic AI deployments for financial oversight.

Google and Southeast Asia's Sea Ltd formed a strategic partnership to develop AI agents for Shopee marketplace and Garena gaming platforms. The collaboration includes an agentic shopping prototype that can assist with purchase decisions and operational processes, reflecting the industry shift toward task-executing agents in commerce.

Cognizant expanded its Google Cloud partnership to help enterprises deploy agentic AI at scale. The partnership includes a dedicated Gemini Enterprise Center of Excellence, Agent Development Lifecycle (ADLC), and Cognizant Agent Foundry for no-code agent development. Focus areas include AI-powered contact centers and intelligent order management systems.

Google launched the Universal Commerce Protocol (UCP) at the National Retail Federation conference, enabling AI agent-based shopping across retailers like Shopify, Etsy, Target, and Walmart. The protocol works with other agent standards including Agent2Agent (A2A) and allows direct checkout from Google Search AI mode results using Google Pay.

Agent Frameworks

LangChain 1.0 and LangGraph 1.0 were released as the first major versions of the open source frameworks, with LangGraph being described as "the first stable major release in the durable agent framework space — a major milestone for production-ready AI systems." After more than a year of iteration and widespread adoption by companies like Uber, LinkedIn, and Klarna, LangGraph is officially v1.

MongoDB introduced the MongoDB Store for LangGraph—a powerful integration that brings flexible and scalable long-term memory to AI agents. This new integration allows agents to remember and build on previous interactions across multiple sessions instead of only retaining memory for the current session. The new langgraph-store-mongodb package introduces a MongoDBStore class, available now through PyPI.

OpenAI announced partnerships with Boston Consulting Group, McKinsey & Co., Accenture, and Capgemini that will see the consulting firms helping sell and implement OpenAI's new Frontier AI agent platform. Frontier is a system that allows businesses and organizations to build, deploy, supervise, and govern AI agents. It is part of OpenAI's effort to seize momentum in the enterprise AI market from its archrival, Anthropic.

OpenAI launched Frontier, a new platform designed for enterprises to build and deploy agents while treating them like human employees. On Thursday, AI giant OpenAI announced the launch of OpenAI Frontier, an end-to-end platform designed for enterprises to build and manage AI agents. It's an open platform, which means users can manage agents built outside of OpenAI too. Frontier users can program AI agents to connect to external data and applications.

The collaboration integrates Anthropic's Claude models and Claude Code with Infosys Topaz to help companies speed up software development and adopt AI with governance and transparency. A core focus of the collaboration is agentic AI—systems that go beyond answering questions to independently handling multi-step tasks. Using tools like the Claude Agent SDK, Infosys and Anthropic will help clients build AI agents that work persistently across long, complex tasks and processes.

The data shows that 60% of organizations run AI agents in production environments. Nearly all describe building agents as a strategic priority. 40% of respondents identify security and compliance as the primary obstacle to scaling agentic AI. 48% identify operational complexity from orchestrating multiple components as the primary challenge in building agents.

Enterprise ANO Adoption

Anthropic unveiled its new enterprise agents program with plug-ins for specific departments like finance, legal, and HR. The launch includes enterprise connectors for Gmail, DocuSign, and Clay, allowing agents to pull data and context directly from linked systems with controlled data flows and centralized administrative controls.

Salesforce reported closing over 22,000 Agentforce deals in Q4, representing nearly 50% quarter-over-quarter growth. Combined annual recurring revenue for Agentforce and Data Cloud reached approximately $1.8 billion, with the platform serving 11.14 trillion tokens in the quarter, indicating production-scale deployment.

Microsoft Agent Framework reached release candidate status, providing a comprehensive open-source framework for building, orchestrating, and deploying AI agents. It's the successor to Semantic Kernel and AutoGen, offering unified programming model across .NET and Python with multi-provider support and workflow orchestration.

Multi-agent workflows grew more than 300% as organizations moved from pilot phases into production. Companies are deploying coordinated digital networks where agents harvest data, validate it, execute transactions, and ensure compliance. The shift represents moving from generative AI to agentic AI focused on execution rather than information retrieval.

Strategy Inc CEO Phong Le declared the traditional era of enterprise software and BI effectively over, being replaced by AI and enterprise sovereignty. The company unveiled Strategy Mosaic as a universal semantic layer and digital twin of organizations that AI can reason over in real-time, with retail agents driving 70% increases in analytics process efficiency.

ANO Security & Governance

NIST launched the AI Agent Standards Initiative through CAISI on February 17, focusing on three pillars: industry-led standards development, open source protocols, and AI agent security research. The initiative includes an RFI on AI Agent Security due March 9, 2026.

Survey of 919 executives reveals 88% of organizations reported AI agent security incidents in the last year. Only 47.1% of AI agents are actively monitored, and just 14.4% have full security approval before deployment.

Veza launched AI Access Agents on February 25, designed to automate identity and access governance tasks. The company also enhanced its AI Agent Security product with deeper visibility into agent risks and MCP server monitoring.

Analysis reveals network layer security gaps in AI agent deployments. NIST's AI Agent Standards Initiative (February 2026) focuses on security frameworks, while a Dark Reading poll found 48% rank agentic AI as the top attack vector for 2026.

New Relic unveiled its Agentic Platform on February 24, a no-code platform for enterprises to build data observability AI agents. The platform supports MCP integration and includes prebuilt agents for monitoring enterprise data systems.

Multi-turn attack success rates reached 92% in testing across eight open models. Only 29% of organizations deploying agentic AI reported being prepared to secure those deployments, creating significant exposure gaps.

Salesforce reported over 22,000 Agentforce deals in Q4 2026, with combined ARR reaching $1.8 billion. The platform processed 11.14 trillion tokens, indicating production-scale enterprise AI agent deployment across Fortune 500 companies.

ANO Academic Research

Multi-agent workflows, within which multiple AI tools jointly automate tasks, saw a 327 percent growth in the Databricks platform. The Supervisor Agent is the most popular, directing and monitoring the actions of other agents, while the Information Extraction agent takes second place.

Nemotron 3 Nano delivers 4x higher throughput than Nemotron 2 Nano and delivers the most tokens per second for multi-agent systems at scale through a breakthrough hybrid mixture-of-experts architecture. The Nemotron 3 models introduce a breakthrough hybrid latent mixture-of-experts (MoE) architecture that helps developers build and deploy reliable multi-agent systems at scale.

Google Research evaluated five architectures and found that parallelizable tasks benefit greatly from multi-agent coordination, with centralized coordination improving performance by 80.9% over a single agent. However, sequential reasoning tasks tend to suffer when multiple agents are introduced, degrading performance by 39-70% due to communication overhead.

While enterprises race to deploy AI agents in customer service, most efforts stall before reaching production, with only 1 in 10 agentic AI pilots making it to live CX environments. All agents are coordinated by an AI Supervisor, which orchestrates sequencing, enforces policy controls, and escalates to human agents when required.

Snowflake and OpenAI announced a multi-year $200 million partnership to make OpenAI models natively available across Snowflake's enterprise data platform, enabling organizations to build agents that reason over governed data. The collaboration positions AI as embedded enterprise infrastructure rather than standalone experimentation.

Most multi-agent AI systems fail expensively before they fail quietly, with Agent A completing a subtask and Agent B reexecuting the same operation with different parameters, causing inconsistent results. The system produces output—but the output costs three times what it should and contains errors that propagate through every downstream task.

ANO Regulation & Policy

NIST announced the launch of the AI Agent Standards Initiative on February 17, 2026, aimed at ensuring AI agents capable of autonomous actions are widely adopted with confidence and can interoperate securely across the digital ecosystem. The initiative focuses on industry-led standards development, open source protocol development, and advancing research in AI agent security and identity.

February 2026 marks a critical month in the battle over AI regulation in America, with California, Texas, Colorado, New York, and Illinois enacting significant AI regulations scheduled to take effect. President Trump's December 2025 executive order declaring federal intent to consolidate AI oversight has created tension with state-level AI laws.

Published February 5, 2026, this analysis covers President Trump's Executive Order signaling federal intent to consolidate AI oversight and new comprehensive governance frameworks in Colorado and California. Companies face a rapidly shifting compliance landscape with AI-specific addenda and third-party risk allocation becoming standard in vendor contracting.

California's Transparency in Frontier AI Act and Texas's Responsible AI Governance Act took effect January 1, 2026, but President Trump's December 11, 2025 Executive Order creates uncertainty. The order establishes an AI litigation task force to challenge state AI laws deemed inconsistent with federal policy and directs commerce secretary to evaluate burdensome state laws by March 11, 2026.

Google Cloud predicts 2026 as the year AI agents "fundamentally reshape business" but only for companies treating them as infrastructure. The shift toward multi-agent systems where specialized agents collaborate on broader tasks is accelerating, with highest-ROI deployments in document processing, data reconciliation, and compliance checks.