xAI & Grok

xAI rolled out Custom Agents capability on March 4, 2026, allowing users to design their own AI agents with distinct personalities and instruction sets. Users can configure up to 4 custom agents through a new 'Your Agents' screen, marking a significant expansion of xAI's multi-agent platform.

xAI's Grok Build CLI agent enables parallel processing using up to 8 concurrent AI agents for code planning, searching, and building. Features local-first architecture with code execution on developer hardware and Arena Mode for comparing agent outputs side-by-side.

xAI released Grok 4.20 Heavy featuring 16 specialized agents that collaborate on complex research projects. Available to SuperGrok Heavy tier users, this version maintains peak performance without limits for demanding workloads.

xAI launched Grok Voice Agent API delivering low-latency speech in multiple languages with tool calling and real-time data access. Includes expressive voice options and OpenAI Realtime API compatibility, deployed across Grok mobile app and Tesla vehicles.

In February 2026, DBS Bank and Visa completed successful tests of AI-driven agentic commerce where software agents autonomously executed credit card transactions. BridgeWise unveiled AI wealth agent capable of personalizing investment portfolios at scale.

Grok 4.20 beta features four cooperating AI agents (Grok, Harper, Benjamin, Luca) processing requests in parallel with 256K token context window. The internal debate model reduces hallucinations, available to SuperGrok subscribers and X Premium+ members.

xAI's Colossus project has deployed 200,000 GPUs with plans to expand to 1 million. This massive infrastructure buildout supports the multi-agent system deployments and positioning xAI as a challenger in AI infrastructure and scale.

OpenAI Agents

OpenAI has released GPT-5.4 on the AgentX platform as of March 9, 2026. This represents the latest version of OpenAI's agent capabilities available for developers on the AgentX platform.

OpenAI secured a $110B funding round valuing the company at $730B. The funding enables enterprises to run OpenAI's agent-based workloads on AWS Bedrock as an alternative to Azure infrastructure.

Meta acquired Moltbook, an AI agent social network that had 2.8 million registered agents by March 2026. This acquisition directly competes with OpenAI's agent ambitions and Google's Gemini agent ecosystem.

OpenAI released guidance on March 2, 2026 for teams developing agentic AI systems, covering AI orchestration patterns and agent development best practices.

Anthropic & Claude

Intuit announced a partnership with Anthropic in March 2026 to integrate the Claude Agent SDK into the Intuit platform, enhancing AI capabilities across their product suite.

Microsoft 365 Copilot now features Anthropic's Claude agent technology in the first direct integration of this model into the platform, marking a significant partnership between the two companies.

Anthropic released a March 2026 update to Skill-creator, its authoring tool for Claude Agent Skills, enabling authors to write evaluations, run benchmarks, and maintain skills with improved automation.

Anthropic announced Auto Mode for Claude Code as a research preview feature expected to roll out no earlier than March 12, 2026, enabling longer autonomous coding sessions.

Microsoft Copilot

Microsoft has expanded model choice capabilities in Microsoft 365 Copilot and Microsoft Copilot Studio to include the GPT-5.3 Instant model. This expansion provides organizations with more AI model options for their Copilot implementations.

Microsoft introduced Copilot Cowork, a new capability that enables users to delegate work to AI by describing desired outcomes. The system converts user requests into automated workplace actions.

Updates include a floating Copilot icon in File Previewer for OneDrive, providing users with a quick, visible entry point to Copilot functionality within file management workflows.

Google & Gemini

Google has expanded its partnership with the Department of Defense to deploy Gemini AI agents for unclassified work. According to Emil Michael, the under secretary of defense for research and engineering, these AI agents will be utilized by U.S. employees for authorized tasks.

The Pentagon has introduced Google AI agents specifically designed to automate non-classified routine tasks for Department of Defense operations. This deployment represents a significant expansion of AI integration within the military.

The Pentagon has officially deployed Google AI agents to handle routine task automation across the Department of Defense. This initiative aims to improve efficiency in unclassified operations.

Google has expanded its existing Gemini partnership with the Department of Defense by introducing specialized AI agents for unclassified projects. The deployment will assist Pentagon employees in their daily operations.

Agent Frameworks

Practical tutorial for building LangGraph agents in Python with a 30-minute timeframe. Highlights LangGraph's memory capabilities through checkpointer functionality for maintaining agent state.

Enterprise ANO Adoption

Joint announcement of a partnership delivering real-time resilience and recovery capabilities for enterprise AI agents. Part of Cohesity's Catalyst on Tour initiative.

Reports on NVIDIA's development of an open-source platform for enterprise AI agents called NemoClaw. Expected to be unveiled at GTC 2026 in March.

Discusses Microsoft's introduction of Microsoft Agent 365, a platform for managing enterprise AI agents. Evaluates the sustainability of AI investment trends.

Identifies enterprise AI agents as the solution poised to be the most widely adopted in 2026, signaling market consensus on agent-centric architectures.

ANO Security & Governance

Major funding round for AI agent security startup founded by Mandiant's founder. Reflects significant venture capital investment in specialized AI agent security solutions for enterprises.

OpenAI's acquisition of Promptfoo addresses critical gaps in AI agent testing and security capabilities. Indicates major platform consolidation around agentic AI security infrastructure.

UiPath achieves AIUC-1 certification for AI agent security and reliability. Establishes industry standards and best practices for agent-native organizations.

Token Security recognized for identity-first approach to AI agent security. Highlights emerging focus on identity and access management as foundational to agent security strategies.