Meta's $2B Manus Acquisition Marks Aggressive Shift From Chat to Agent-Driven Action

In partnership with

The Latest Agentic Development

Meta Acquires Manus for Autonomous Agent Leadership

Meta has acquired Manus, a Singapore-based developer of general-purpose AI agents, in a move to embed autonomous task execution across its consumer and business products. Manus reached $100 million in annual recurring revenue within 8 months of launch the fastest any startup has achieved this milestone—with a revenue run rate exceeding $125 million. The platform has processed over 147 trillion tokens and created 80 million virtual computers, placing it among the most heavily used agent systems globally. Meta will integrate Manus's agent capabilities directly into Facebook, Instagram, WhatsApp, and its business products while maintaining the service as a standalone offering.​

Strategic Implication: This acquisition signals Meta's commitment to building a complete agent ecosystem across its 3+ billion-user platform, positioning it to compete directly with OpenAI, Google, and Microsoft in enterprise agentic AI while maintaining consumer distribution advantages.

Machine Identities Outnumber Humans 82-to-1; IAM Crisis Looms

Research from CyberArk reveals machine identities now outnumber humans 82 to 1, with AI agents representing the fastest-expanding and least-regulated segment. Microsoft Copilot Studio users generated over 1 million AI agents in a single quarter a 130% increase year-over-year. ServiceNow invested $11.6 billion in security acquisitions in 2025 alone, signaling a strategic pivot toward AI identity and risk management. Gartner forecasts that by year-end 2025, 8.25% of enterprise security breaches will be linked to AI agent misuse, exposing a critical governance gap in human-centric identity frameworks designed when machines were exceptions, not the norm.​

Strategic Implication: The identity sprawl created by agent proliferation is forcing enterprises to rethink security architecture. Companies investing in machine identity governance, endpoint visibility, and agentic observability will establish competitive moats in enterprise AI deployment.

Google's AP2 Protocol Competes With OpenAI on Agentic Checkout

Google has scaled its agentic commerce infrastructure with "Agentic Checkout," introduced in November 2025, enabling users to set "Price Mandates" while AI agents negotiate and execute transactions in real-time. The Google Shopping Graph now indexes over 50 billion product listings with 2 billion updates per hour—real-time product data that ensures AI agents operate on the most current inventory and pricing information. Google's AP2 protocol now competes directly with OpenAI's Agentic Commerce Protocol (ACP), with both platforms targeting merchants and platform operators. The expansion signals a fundamental shift in consumer behavior from "search-and-click" to "intent-and-delegate."​

Strategic Implication: Google's real-time product indexing and native agentic checkout create a data advantage that enables AI agents to execute complex multi-step transactions (price comparison, inventory verification, deal closure) without merchants' direct involvement. This represents a disintermediation risk for traditional e-commerce players.

Microsoft Foundry Agent Service Adds Managed Memory for Long-Term Agent State

Microsoft announced that Foundry Agent Service now includes managed, long-term memory a state layer for agentic systems. Developers can store, retrieve, and manage chat summaries, user preferences, and critical context across sessions, devices, and workflows without custom application code. This shift moves memory management from application logic to service infrastructure. During public preview, the feature is free; users pay only for underlying model usage. Current constraints: up to 10,000 discrete memory items per scope, 1,000 requests/minute throughput. This represents a fundamental architectural shift, turning long-term context from a custom implementation to an enterprise primitive.​

Strategic Implication: Native agent memory eliminates a core engineering challenge for enterprise deployments. Agents can now maintain persistent state across touchpoints, enabling complex multi-turn workflows and personalization at scale. This is critical infrastructure for agents managing customer journeys, compliance records, or multi-step business processes.

WineSpeak.ai debuts a DTC “AI Concierge” agent for winery commerce funnels

A PRNewswire release spotlights Fogarty Winery deploying a custom AI concierge agent (“Thomas”) to guide visitors through experiences, membership, and wine selection. While vertical, it’s a clean example of agentic commerce moving from generic chatbots to conversion-tied flows (curation → recommendation → purchase intent) embedded directly into web funnels.

What to Watch this Week

 🎙️AI on Fire: Real Builders. Real Heat.

Stories from founders building AI Agents. Explore all episodes here

⚔️ AArena: The Battleground for AI

Stop demo-hopping. One workspace. Every agent. Real results.

👉 Enter the Arena | 👉 See the Battleboard

🏆 The Leaderboard Never Sleeps

The global ranking of AI agents is shifting every day. Who’s on top? Who just dropped?

👉 Explore the Leaderboard →

🗺️ The Map of AI Agents (Live & Growing)

We’re charting the entire AI agent ecosystem — thousands of options across categories.
Your next agent is already on the map.

👉 Explore the Landscape Map →