#12: The AI Agent Gold Rush

Mistral launches Agent API. Microsoft introduced CoreAI. IBM launches HTTP for AI agents. Capgemini's Agentic AI Whitepaper. And much more.

Marc Baumann

and

Sangam Bharti

Jun 01, 2025

∙ Paid

Hey, it’s Marc!

AI Agents are moving faster than you are thinking.

One moment you’re optimising your strategy. The next, it’s irrelevant.

“GenAI is not just another tech cycle—it is a civilizational shift. The path is clear: IT and business services are moving toward autonomous operations. The rise of autonomous robots and AI agents promises a future of ‘dark factories’ and AI-assisted enterprise functions.”
— N Chandrasekaran, Chairman, TCS

Here are the top Agent signals:

Microsoft introduced CoreAI, an all-in-one toolkit and cloud foundation for developers.
Anthropic introduced Claude 4, two AI power modes—Opus 4 for deep, continuous tasks and Sonnet 4 for instant reasoning—and now keeps your context alive for up to an hour.
Capgemini introduced an Agentic AI whitepaper to understand and implement autonomous AI agent systems in their organisations.

Power Moves

Mistral AI launches Agents API for enterprise automation

Mistral AI just rolled out its Agents API—one SDK that bundles server-side conversation management, a Python code interpreter, web search (AFP/AP tiers), FLUX1.1 image generation, document retrieval, multi-agent orchestration and Model Context Protocol support—so developers can build fully autonomous, multistep AI agents without stitching together disparate tools. With dozens of pre-built connectors, enterprises can now deploy agentic workflows for tasks like financial analysis, report generation and ticket triage—cutting integration overhead by over 50% and slashing time-to-market. ANNOUNCEMENT

Infosys ships 200+ AI agents on Google Cloud

Infosys has launched over 200 ready-to-use AI agents via its Topaz suite on Google Cloud’s Vertex AI—autonomous bots for finance, healthcare, manufacturing, retail and more that handle everything from network monitoring to demand forecasting—so enterprises can deploy multistep, secure workflows without custom integrations. By bundling real-time decision-making, data extraction and multimodal processing into prebuilt agents, companies can cut automation rollout time by weeks and scale human+AI collaboration across departments. ANNOUNCEMENT

IBM launches HTTP for AI agents

IBM just released the Agent Communication Protocol (ACP), a RESTful, HTTP-based standard—already powering the open-source BeeAI platform under the Linux Foundation—that lets LLM agents talk peer-to-peer and orchestrate complex, multi-agent workflows without brittle, custom integrations; when paired with MCP for tool access, enterprises can deploy decentralized AI pipelines out of the box. RELEASE

Vendor Spotlight: AlphaEvolve

What is AlphaEvolve? An autonomous coding agent that uses advanced LLMs in an evolutionary loop to optimise and invent algorithms – directly editing code, evaluating outcomes, and iterating without manual intervention.

Core Business Functions:

Compute & Resource Optimisation: Automatically fine-tunes how workloads are scheduled and balanced across servers to maximise utilisation
AI/ML Kernel Acceleration: Evolves low-level code for critical ML operations—think matrix multiplies or attention mechanisms
DevOps & CI/CD Tuning: Optimises build scripts, test pipelines, and deployment flows by evolving configuration files and commands

The Payoff

Lower costs
Faster results
New growth opportunities

More on AlphaEvolve:

👉 Want the spotlight? Get your brand in front of 30,000+ decision-makers — book your ad spot now.

Top AI Vendors This Week

Flow: AI-driven filmmaking platform that lets creators quickly craft and refine cinematic scenes using simple, natural language
Jules: AI coding partner that quietly handles tasks like writing tests, fixing bugs, and building features in the codebase
Clark: AI agent that builds secure, production-ready internal enterprise apps
Catena Labs: AI-native financial infrastructure and service provider