News from the AI & ML world

DeeperML - #aiagents

Kuldeep Jha@Verdict //
Databricks has unveiled Agent Bricks, a new tool designed to streamline the development and deployment of enterprise AI agents. Built on Databricks' Mosaic AI platform, Agent Bricks automates the optimization and evaluation of these agents, addressing the common challenges that prevent many AI projects from reaching production. The tool utilizes large language models (LLMs) as "judges" to assess the reliability of task-specific agents, eliminating manual processes that are often slow, inconsistent, and difficult to scale. Jonathan Frankle, chief AI scientist of Databricks Inc., described Agent Bricks as a generalization of the best practices and techniques observed across various verticals, reflecting how Databricks believes agents should be built.

Agent Bricks originated from the need of Databricks' customers to effectively evaluate their AI agents. Ensuring reliability involves defining clear criteria and practices for comparing agent performance. According to Frankle, AI's inherent unpredictability makes LLM judges crucial for determining when an agent is functioning correctly. This requires ensuring that the LLM judge understands the intended purpose and measurement criteria, essentially aligning the LLM's judgment with that of a human judge. The goal is to create a scaled reinforcement learning system where judges can train an agent to behave as developers intend, reducing the reliance on manually labeled data.

Databricks' new features aim to simplify AI development by using AI to build agents and the pipelines that feed them. Fueled by user feedback, these features include a framework for automating agent building and a no-code interface for creating pipelines for applications. Kevin Petrie, an analyst at BARC U.S., noted that these announcements help Databricks users apply AI and GenAI applications to their proprietary data sets, thereby gaining a competitive advantage. Agent Bricks is currently in beta testing and helps users avoid the trap of "vibe coding" by forcing rigorous testing and evaluation until the model is extremely reliable.

Recommended read:
References :
  • www.bigdatawire.com: Databricks Wants to Take the Pain Out of Building, Deploying AI Agents with Bricks
  • siliconangle.com: The best judge of artificial intelligence could be AI — at least that’s the idea behind Databricks Inc.’s new tool, Agent Bricks.
  • thenewstack.io: Databricks Launches Agent Bricks, Its New No-Code AI Agent Builder
  • www.infoworld.com: Databricks has released a beta version of a new agent building interface to help enterprises automate and optimize the agent building process.
  • thenewstack.io: Databricks Launches Agent Bricks, Its New No-Code AI Agent Builder
  • AI News | VentureBeat: Databricks Agent Bricks automates enterprise AI agent optimization and evaluation, eliminating manual processes that block production deployments.
  • SiliconANGLE: The best judge of artificial intelligence could be AI — at least that’s the idea behind Databricks Inc.’s new tool, Agent Bricks.
  • BigDATAwire: Databricks today launched Agent Bricks, a new offering aimed at helping customers AI agent systems up and running quickly, with the cost, safety, and efficiency they demand.
  • Analytics India Magazine: Databricks also launched MLflow 3.0, a redesigned version of its AI lifecycle management platform.
  • Verdict: Databricks introduces Agent Bricks for AI agent development
  • www.verdict.co.uk: Databricks introduces Agent Bricks for AI agent development
  • www.bigdatawire.com: Databricks Is Making a Long-Term Play to Fix AI’s Biggest Constraint
  • techstrong.ai: Highlights Databricks' simplified approach to building and training AI agents.
  • siliconangle.com: Reveals Databricks' play for AI agents and their data platform strategy.
  • techstrong.ai: Databricks this week launched a series of initiatives, including a beta release of an Agent Bricks framework that makes it simpler to create and modify artificial intelligence agents using techniques developed by Mosaic AI Research using multiple types of large language models (LLMs).

@thenewstack.io //
Emerging tools are revolutionizing AI agent development and management. Databricks recently launched Agent Bricks, a no-code AI agent builder, simplifying the creation process. Complementing this, Google's Gemini Agent Network Protocol offers a framework for intelligent collaboration among AI agents, enabling dynamic communication and task distribution. These advancements signify a move toward more accessible and collaborative AI agent ecosystems.

Vanta has introduced an AI agent designed to automate security compliance workflows, promising to save enterprises significant time on policy management and audit preparation. This agent proactively identifies compliance issues, suggests fixes, and takes action while keeping humans in control. By minimizing human error and automating repetitive tasks, Vanta's AI agent allows security teams to focus on higher-value work, addressing the increasing time organizations spend on compliance, particularly as security risks escalate.

The Vanta AI Agent addresses critical areas that typically consume hundreds of hours of manual work. It automates policy onboarding by scanning documents, extracting key details, and mapping policies to relevant compliance controls. This eliminates bottlenecks associated with manual control mapping and generates policy change summaries, streamlining annual reviews. The use of Gemini models by Google in the Gemini Agent Network Protocol enables automated task distribution, collaborative problem-solving, and enriched dialogue management, making it ideal for complex data analysis and information validation.

Recommended read:
References :
  • thenewstack.io: Databricks Launches Agent Bricks, Its New No-Code AI Agent Builder
  • venturebeat.com: Vanta’s AI agent wants to run your compliance program — and it just might
  • www.marktechpost.com: How to Build an Asynchronous AI Agent Network Using Gemini for Research, Analysis, and Validation Tasks

@learn.aisingapore.org //
AI agents are rapidly transitioning from simple assistants to active participants in enterprise operations. This shift promises to revolutionize workflows and unlock new efficiencies. However, this move towards greater autonomy also introduces significant security concerns, as these agents increasingly handle sensitive data and interact with critical systems. Companies are now grappling with the need to balance the potential benefits of AI agents with the imperative of safeguarding their digital assets.

The Model Context Protocol (MCP) is emerging as a key standard to address these challenges, aiming to provide a secure and scalable framework for deploying AI agents within enterprises. Additionally, the concept of "agentic security" is gaining traction, with companies like Impart Security developing AI-driven solutions to defend against sophisticated cyberattacks. These solutions leverage AI to proactively identify and respond to threats in real-time, offering a more dynamic and adaptive approach to security compared to traditional methods. The complexity of modern digital environments, driven by APIs and microservices, necessitates these advanced security measures.

Despite the enthusiasm for AI agents, a recent survey indicates that many organizations are struggling to keep pace with the security implications. A significant percentage of IT professionals express concerns about the growing security risks associated with AI agents, with visibility into agent data access remaining a primary challenge. Many companies lack clear policies for governing AI agent behavior, leading to instances of unauthorized system access and data breaches. This highlights the urgent need for comprehensive security strategies and robust monitoring mechanisms to ensure the safe and responsible deployment of AI agents in the enterprise.

Recommended read:
References :
  • orases.com: Organizational leaders are entering a period where autonomous AI agents are poised to dramatically change how enterprises operate at scale.
  • AI News | VentureBeat: AI agents are moving from passive assistants to active participants. Today, we ask them to do. Tomorrow, we’ll authorize them to act.
  • thenewstack.io: Deploying A Secure Enterprise Agentic AI: MCP + Agent2Agent
  • www.techradar.com: Love and hate: tech pros overwhelmingly like AI agents but view them as a growing security risk
  • AI Accelerator Institute: Agents of change or agents of chaos?
  • composio.dev: MCP agents can now interact with real apps and accomplish tasks.
  • siliconangle.com: Amplitude launches AI Agents to speed up product decision-making

Jesus Rodriguez@TheSequence //
References: CustomGPT , TheSequence ,
Advancements in AI agent development are rapidly transforming how organizations access data and automate tasks. Custom AI agents are emerging as a powerful tool, offering domain-specific responses and actions that make interactions more intuitive and effective. These agents are purpose-built, leveraging domain-specific fine-tuning to align with unique operational needs, unlike general AI models that serve broad purposes. Companies are finding that these custom agents handle niche queries and complex workflows with greater precision, leading to significant improvements in efficiency and accuracy.

Custom AI agents enable organizations to access data and automate tasks with tailored responses, making interactions intuitive and effective. Building these agents involves a series of steps, from gathering relevant domain data and defining precise objectives to selecting or fine-tuning a foundation model and designing conversational flows. As you build your agent, you’ll iterate on design, test performance, and refine responses so it meets requirements and adapts to evolving needs. Techniques like semantic indexing and entity recognition ensure the agent understands relationships between concepts, improving its ability to retrieve and process information.

Partnering is also allowing companies to Orchestrate large-scale agent training. Reasoning agents are among the most sought-after LLM use cases, automating complex tasks across domains. With Lambda’s 1-Click Clusters and dstack’s orchestration, teams spend less time on setup and more on building. Self-improving agents can rewrite their own code to enhance performance. Built atop frozen foundation models, these agents alternate between self-modification and evaluation, benchmarking candidate agents on real-world coding tasks.

Recommended read:
References :
  • CustomGPT: A custom AI agent changes how organizations access data and automate tasks by providing domain-specific responses and actions, making interactions more intuitive and effective.
  • TheSequence: Agents that improve themselves and the limits of memorization.
  • AI Accelerator Institute: What is an AI agent? Learn how to build them, how to scale them, and why most teams never make it past the prototype phase.

Justin Westcott,@AI News | VentureBeat //
AI agents are poised to revolutionize how we interact with the internet, moving beyond passive assistants to active participants authorized to act on our behalf. This shift necessitates a redesign of the web, transforming it from a human-centric interface to a machine-native environment optimized for speed, efficiency, and transactional capabilities. The current internet, designed for human eyes and fingers, is proving inefficient for AI, which requires structured data, clear intent, and exposed capabilities to navigate, decide, and transact effectively. This evolution will lead to a web where APIs become the new storefronts, prioritizing verifiable sources and trust over traditional user experience elements.

The development and deployment of AI agents face significant challenges, particularly in ensuring reliability and consistency within defined business processes. Existing agentic frameworks often fall short due to a lack of state, leading to unpredictable behavior and poor adherence to workflows. A recent survey highlighted that only 25% of AI initiatives are live in production, with hallucinations and prompt management being major obstacles. This indicates a need for robust evaluation processes and automated testing pipelines to address these issues, as traditional software QA methods may not fully apply to AI applications. The survey indicated that without robust evaluation, AI agents may not reach production or may not be sustainable long term.

An alternative approach, known as process calling, aims to create reliable, process-aware, and easily debuggable conversational agents. This method addresses the limitations of tool calling by incorporating state tracking and structured workflows. Companies achieving success with LLMs are prioritizing robust evaluation and moving beyond simple tool-based interactions. As AI agents become more prevalent, the internet will likely bifurcate into two webs: one designed for humans and another designed for machines. This machine-native web will feature faster protocols, cleaner metadata, and a focus on verifiable sources, ultimately reshaping the architecture of the internet to accommodate AI's unique requirements.

Recommended read:
References :

Alexey Shabanov@TestingCatalog //
AI agents are rapidly transforming how work gets done by automating and streamlining a variety of workflows. These intelligent systems are designed to handle tasks ranging from managing schedules, emails, and notes, as exemplified by Genspark's new AI Secretary feature, to providing personalized customer engagement in the automotive retail sector, demonstrated by Impel's use of fine-tuned LLMs. The core advantage of agentic AI lies in its capacity for autonomous decision-making and enhanced customer experiences powered by AI-driven solutions. Impel, for instance, optimizes automotive retail customer connections through personalized experiences at every touchpoint, utilizing Sales AI to provide instant responses and maintain engagement during the car-buying journey.

The development of agentic AI extends to the realm of IoT, where these agents are poised to enable autonomous, goal-driven decision-making. This is particularly relevant in smart homes, cities, and industrial systems, where AI agents can proactively address network issues, strengthen security, and improve overall productivity. Agentic AI marks a structural shift from traditional AI, transitioning from task-specific and supervised models to autonomous agents capable of real-time decisions and adaptation. These agents possess memory, autonomy, task awareness, learning, and reasoning abilities, allowing them to operate with minimal human intervention.

However, the effectiveness of AI agents hinges on accurate monitoring strategies and their ability to navigate complex tasks. To ensure reliability in real-world scenarios, benchmarks like WebChoreArena are being developed to challenge agents with memory-intensive and reasoning-intensive scenarios. Building robust conversational AI agents also requires overcoming limitations in existing frameworks. The Rasa platform offers an alternative approach through process calling, enabling the creation of reliable, process-aware, and easily debuggable conversational agents. This method addresses issues such as loss of conversational context and poor adherence to business processes, ensuring that AI agents can consistently guide users through predetermined workflows.

Recommended read:
References :

@Salesforce //
References: Salesforce
The modern workplace is undergoing a significant transformation with the integration of AI agents into daily operations. Organizations are increasingly adopting autonomous AI systems capable of automating entire workflows across various industries. This shift marks the beginning of the "agentic AI era," where intelligent systems can perform complex tasks, make decisions, and interact with systems with minimal human oversight. This evolution requires organizational leaders to strategically plan for AI agent integration and implementation to maximize efficiency and productivity.

This new collaborative workforce sees humans and AI agents working together, fundamentally changing roles, workflows, and strategies. The contact center is a prime example, moving away from a "bot vs. human" approach to a "bot with human" model. In this landscape, human agents become orchestrators of complex customer journeys, while AI agents act as autonomous copilots, taking initiative on routine tasks and deferring to humans when emotional intelligence or nuanced understanding is required. This collaborative approach aims to improve speed, operational throughput, and decision quality.

Companies such as Cisco are building the infrastructure to support this AI-driven future, where potentially billions of AI agents will work together globally and continuously. This includes developing systems where AI agents can independently handle tasks such as identity verification, multi-step backend actions, and even proactive customer engagement. However, successful integration requires careful consideration of data integration, system compatibility, governance, compliance, and change management to ensure AI agents operate within predefined boundaries and judgment frameworks.

Recommended read:
References :
  • Salesforce: The New Collaborative Workforce: Humans and Digital Agents

@orases.com //
References: www.marktechpost.com , Orases , Maginative ...
AI agents are rapidly transforming industries by automating tasks and enhancing decision-making, moving beyond simple automation to intelligent autonomy. These agents are being implemented across various sectors, promising significant improvements in efficiency and productivity. A strategic roadmap is essential for successful AI agent implementation, aligning technology with workflows and business objectives to ensure that these systems have a real impact on operations and decision-making. Without a clear structure, companies risk wasting investments on generic tools and isolated pilot projects.

The impact of AI agents is particularly evident in customer experience (CX), with companies increasingly integrating AI agents into their technology interactions. Cisco's recent Agentic AI Report highlights the transformative impact of these autonomous agents, which can retain memory, reason about tasks, and autonomously select actions to optimize outcomes with minimal human intervention. Cisco's data anticipates that enterprises expect 56% of their interactions with technology partners will be managed by AI agents within the next 12 months, increasing to 68% over three years. This accelerated adoption necessitates that vendors rapidly develop and deploy scalable, robust agentic AI solutions.

Thomson Reuters is also leveraging this trend with agentic AI capabilities in its CoCounsel assistant, enabling autonomous, multi-step task execution in tax and accounting workflows. Early results show that processes like tax jurisdiction reviews have been drastically reduced from half a week to under an hour. The company plans to extend agentic AI to legal, risk, and compliance domains, connecting firm knowledge, codes, and internal documents into one workspace where AI handles complete workflows, rather than individual queries. This integration allows professionals to focus on higher-level tasks, ensuring that human expertise guides judgment and validates outputs.

Recommended read:
References :
  • www.marktechpost.com: Cisco’s Latest AI Agents Report Details the Transformative Impact of Agentic AI on Customer Experience
  • Orases: The Roadmap to Successful AI Agent Implementation
  • www.analyticsvidhya.com: 8 Things to Keep in Mind while Building AI Agents
  • Maginative: Thomson Reuters Adds Agentic Capabilities to CoCounsel

@www.marktechpost.com //
The development of AI agents capable of performing human tasks on computers is gaining momentum, with a particular focus on multi-agent communication systems. Several research labs and companies are actively exploring this area, aiming to build agents that can effectively coordinate and collaborate. A key aspect of this research involves establishing robust communication protocols that enable seamless interaction between multiple AI agents. Recent articles highlight the progress being made in constructing code using these multi-agent communication systems, paving the way for more sophisticated and autonomous AI applications.

Mistral AI recently released its Agents API, providing public access through La Plateforme for developers to create autonomous agents. This API allows agents to plan tasks, utilize external tools, and maintain long-term context. The interface comes equipped with connectors for Python execution, web search, Flux 1.1 image generation, and a document library. The Agents API supports the mistral-medium-latest and mistral-large-latest models, allowing agents to delegate subtasks to each other via the Model Context Protocol, creating coordinated workflows across multiple services.

A tutorial was recently released which provides a coding guide to building scalable multi-agent communication systems using the Agent Communication Protocol (ACP). This guide implements ACP by building a flexible messaging system in Python, leveraging Google's Gemini API for natural language processing. The tutorial details the installation and configuration of the google-generativeai library, introduces core abstractions, message types, performatives, and the ACPMessage data class for standardizing inter-agent communication. Through ACPAgent and ACPMessageBroker classes, the guide demonstrates how to create, send, route, and process structured messages among multiple autonomous agents, also showing how to implement querying, requesting actions, broadcasting information, maintaining conversation threads, acknowledgments, and error handling.

Recommended read:
References :

Priyansh Khodiyar@CustomGPT //
References: CustomGPT , hackernoon.com ,
The Model Context Protocol (MCP) is gaining momentum as a key framework for standardizing interactions between AI agents and various applications. Developed initially by Anthropic, MCP aims to provide a universal method for AI models to connect with external tools, data sources, and systems, similar to how USB-C streamlines connections for devices. Microsoft is actively embracing this protocol, introducing MCP servers for its Dynamics 365 platform. Furthermore, companies are integrating MCP into their APIs, indicating a widespread movement towards its adoption.

The core challenge MCP addresses is the current fragmented and inconsistent nature of AI integrations. Without a standardized protocol, developers often resort to custom code and brittle integrations, leading to systems that are difficult to maintain and scale. MCP standardizes how context is defined, passed, and validated, ensuring that AI agents receive the correct information in the right format, regardless of the data source. This standardization promises to alleviate the "It Works on My Machine… Sometimes" syndrome, where AI applications function inconsistently across different environments.

MCP's adoption is expected to pave the way for more autonomous enterprises and smarter systems. Microsoft envisions a future where AI agents proactively identify problems, suggest solutions, and maintain context across conversations, thereby transforming workflows across diverse fields such as marketing and software engineering. The evolution of identity standards, particularly OAuth, is crucial to secure agent access across connected systems, ensuring a robust and reliable ecosystem for AI agent interactions. This collaborative effort to build standards will empower the next generation of AI agents to operate effectively and securely.

Recommended read:
References :
  • CustomGPT: Problems MCP Model Context Protocol solves
  • hackernoon.com: AI Agents, MCP Protocols, and the Future of Smart Systems
  • www.madrona.com: The End of Biz Apps? AI, Agility, and The Agent-Native Enterprise from Microsoft CVP Charles Lamanna

Coral Garnick@Madrona //
References: AIwire , insideAI News ,
Microsoft is intensifying its efforts in AI agent technology, signaling a significant shift towards the "autonomous enterprise." At Microsoft Build 2025, the company highlighted its focus on agentic AI, including the expansion of Azure AI Foundry to support complex workflows and multi-agent systems. This platform now processes billions of enterprise queries daily, serving over 70,000 customers, and is being enhanced with Grok models from xAI, giving developers a wider selection of AI tools. CVP Charles Lamanna emphasizes that AI agents are fundamentally reshaping teams, tools, and workflows, predicting a future where "there will now be ‘an agent for that’" instead of just an app.

The introduction of the Model Context Protocol (MCP) servers for Microsoft Dynamics 365 ERP and CRM is aimed at enhancing AI agent interoperability. These servers break down data and application silos, allowing agents to work across processes, thereby enabling new autonomous scenarios for improved business functionality and productivity. Microsoft is also pushing for multi-agent workflows and on-device AI through Foundry Local, supporting open protocols like Agent-to-Agent (A2A) and MCP to streamline AI system communication. These protocols are crucial for creating an 'open agentic web', allowing diverse AI systems to work together efficiently.

To ensure practical implementation for developers, Microsoft has introduced AgentOps tools within Azure AI Foundry. These tools provide real-time insights into agent performance, covering aspects such as efficiency, cost, and safety, which are often overlooked during deployment. NVIDIA, through its AI-Q Blueprint, is also addressing the challenge of integrating multiple vendor agent systems by promoting interoperability among agents, tools, and data sources. As Bartley Richardson from NVIDIA noted, enterprises must acknowledge the reality of multi-vendor environments and strive to seamlessly mesh different agent systems to drive efficiency across industries, even if the agentic systems don’t get it right 100% of the time.

Recommended read:
References :
  • AIwire: Details on Microsoft's vision for an 'open agentic web' and its focus on agentic AI at Build 2025.
  • insideAI News: This article discusses the vision of an "open agentic web" from Microsoft, highlighting the significance of developers in AI and the company's embrace of diverse models.
  • www.infoq.com: Microsoft Announces AI Agent and Platform Updates at Build 2025

Alex Simons@Microsoft Security Blog //
Microsoft is aggressively pursuing advancements in AI agent technology, with a focus on secure access and collaborative capabilities. The company's efforts are highlighted by the development of Magentic-UI, an open-source research prototype designed as a human-centered web agent. This agent aims to facilitate real-time collaboration on complex, web-based tasks. Microsoft envisions that within the next two years, AI agents will evolve from simply responding to requests to proactively identifying problems, suggesting solutions, and maintaining context across conversations.

The key to this evolution lies in adapting identity standards, specifically OAuth, to ensure secure agent access to data and systems. Microsoft is building a robust agent ecosystem, including sophisticated elements like MCP servers for Dynamics 365, to enhance AI interaction across various platforms. Magentic-UI, built on the Magentic-One architecture and powered by AutoGen, allows users to directly modify the agent's plan and provide feedback, ensuring transparency and control. It is integrated with Azure AI Foundry models and agents.

Magentic-UI is engineered to support intricate, multi-step workflows through human-AI collaboration, showcasing potential advancements in agentic user experience (AUX). It can perform tasks that involve browsing the web, executing code, and understanding files. Microsoft believes the next generation of AI agents will augment and amplify organizational capabilities, enabling autonomous tasks such as creating marketing campaign plans and developing new software features with minimal human interaction.

Recommended read:
References :

@www.marktechpost.com //
AI agents are rapidly transforming software engineering workflows, offering increased efficiency and accessibility. Mistral AI has launched its Agents API, a platform designed to enable developers to integrate autonomous, generative AI capabilities into existing applications. This API allows for the creation of AI agents capable of performing tasks such as running Python code securely, generating images, and performing retrieval-augmented generation (RAG). These agents can access real-time information from the web and utilize user-provided document libraries, significantly enhancing their ability to provide accurate and up-to-date responses.

Designed to complement Mistral’s existing Chat Completion API, the Agents API focuses on agentic orchestration, built-in connectors, and persistent memory. The API is equipped with several built-in connectors, including Code Execution, Image Generation, Document Library, and Web Search. This flexibility allows for the coordination of multiple AI agents to tackle complex tasks, surpassing the limitations of traditional language models by enabling them to perform real-world tasks and maintain conversational context over time.

The rise of AI agents is also changing the economics of software engineering. The emergence of "cheap SWE agents" is enabling teams with more millions in ARR than employees. Tools like GitHub Copilot, Claude Code, and OpenAI Codex are democratizing the field, making software development accessible to individuals without a technical background. These agents are also improving developer productivity and code quality.

Recommended read:
References :
  • AI News | VentureBeat: Mistral launches API for building AI agents that run Python, generate images, perform RAG and more
  • www.marktechpost.com: Mistral Launches Agents API: A New Platform for Developer-Friendly AI Agent Creation

Alexey Shabanov@TestingCatalog //
Mistral AI is expanding its AI capabilities with the introduction of a new Agents feature within Le Chat, offering users intuitive customization, advanced controls, and faster performance. This redesigned Agents feature replaces the earlier Agent Builder interface and integrates closely with the main chat experience. It allows users to create and customize autonomous agents with functionalities similar to OpenAI's GPT Builder but with its unique design choices and system integrations.

Mistral AI has also launched its Agents API, a framework designed to empower developers to build AI agents capable of executing various tasks. These tasks include running Python code in a secure sandbox, generating images using the FLUX model, and performing retrieval-augmented generation (RAG). The Agents API provides a cohesive environment for large language models to interact with multiple tools and data sources, fostering efficient and versatile AI agent creation.

The features that are converging across major LLM API vendors are code execution (Python in a sandbox), web search (using Brave), document library (hosted RAG), and image generation (FLUX for Mistral). The rate of MCP support is also similar across the major vendors with OpenAI adding it May 21st, Anthropic launched theirs May 22nd and now Mistral has launched theirs on May 27th. For professionals like Lead AI Engineers or Senior AI Engineers, the Mistral Agents API represents a powerful addition to their AI toolkit.

Recommended read:
References :
  • MarkTechPost: Mistral has introduced its Agents API, a framework designed to facilitate the development of AI agents capable of executing a variety of tasks including running Python code, generating images, and performing retrieval-augmented generation (RAG).
  • TestingCatalog: Discover Mistral AI's new Agents feature in Le Chat, offering intuitive customisation, advanced controls, and faster performance!
  • AI News | VentureBeat: For professionals like the Lead AI Engineer or Senior AI Engineer, the Mistral Agents API represents a powerful addition to their AI toolkit.
  • Simon Willison's Weblog: Big upgrade to Mistral's API this morning: they've announced a new "Agents API".
  • www.infoworld.com: Artificial intelligence startup Mistral AI Tuesday announced the , a complement to its that “simplifies implementing agentic use cases,” the company said.
  • Simon Willison: It's interesting how the major LLM API vendors are converging on the following features: - Code execution: Python in a sandbox - Web search - like Anthropic, Mistral seem to use Brave - Document library aka hosted RAG - Image generation (FLUX for Mistral) - MCP MIstral today: The rate MCP support rolled out in the major vendor APIs is pretty astonishing: OpenAI added it May 21st, Anthropic launched theirs May 22nd and now Mistral have launched theirs on May 27th!
  • www.marktechpost.com: Mistral Launches Agents API: A New Platform for Developer-Friendly AI Agent Creation
  • simonwillison.net: Simon Willison
  • TestingCatalog: Mistral AI opens Agents API for public use with task planning and tool integration
  • Artificial Intelligence: Article describing how to build an agentic RAG application using LlamaIndex and Mistral in Amazon Bedrock.
  • www.producthunt.com: Discussion on Mistral's Agents API and its features for building AI agents.

Ken Yeung@Ken Yeung //
Microsoft is making significant strides in AI innovation, with a focus on both experimental and practical applications. One notable project unveiled at Build 2025 is Project Amelie, an experimental AI agent designed to autonomously build machine learning pipelines from a single prompt. Powered by Microsoft Research's RD agent, Amelie aims to automate and optimize research and development processes in machine learning, potentially eliminating manual setup work typically handled by data scientists. Early testing has shown promising results, with Project Amelie outperforming current state-of-the-art benchmarks on MLE-Bench.

Microsoft is also applying AI to solve real-world problems in healthcare and weather forecasting. They have unveiled an AI-powered orchestration system available through the Azure AI Foundry Agent Catalog to streamline cancer care planning, which brings together specialized AI agents to assist clinicians with analyzing multimodal medical data from imaging and genomics to clinical notes and pathology. This system aims to automate parts of the tumor board process, making personalized treatment plans more accessible. In weather forecasting, Microsoft's latest AI model, Aurora, is able to provide detailed and accurate 10-day forecasts in seconds.

In addition to these innovations, Microsoft is advancing its Windows AI strategy with native support for Model Context Protocol (MCP) on Windows 11 and the introduction of Windows AI Foundry. The MCP integration will bring Anthropic's protocol to Windows 11, enabling AI agents to connect with native apps, system services, and external tools. With its Windows AI Foundry, developers can fine-tune and run AI models directly on Windows PCs. These efforts aim to build a secure agentic future on Windows, fostering the development of AI agents within the Windows ecosystem.

Recommended read:
References :
  • AIwire: Microsoft Unveils Agentic AI Tool to Streamline Cancer Care
  • www.windowscentral.com: Microsoft's latest AI model can accurately forecast the weather: “It doesn’t know the laws of physics, so it could make up something completely crazyâ€
  • Ken Yeung: Microsoft’s Project Amelie Is an Experiment in ‘AI Developing AI’
  • Microsoft Research: Magentic-UI, an experimental human-centered web agent
  • learn.aisingapore.org: Code Agents: The Future of Agentic AI
  • www.windowscentral.com: "I’ve got 30% of my time back": Microsoft Copilot reportedly helps execs quickly catch up on work after vacations

@www.marktechpost.com //
Advancements in AI are rapidly shifting towards multi-agent systems, where specialized AI agents collaborate to perform complex tasks. These agents, envisioned as a team of expert colleagues, are designed to analyze data, interact with customers, and manage logistics, among other functions. The challenge lies in orchestrating these independent agents to work together seamlessly, ensuring they can coordinate interactions, manage shared knowledge, and handle potential failures effectively. Solid architectural blueprints are crucial for building reliable and scalable multi-agent systems, emphasizing the need for patterns designed for reliability and scale from the outset.

LangGraph Platform is emerging as a key tool for deploying these complex, long-running, and stateful AI agents. It addresses challenges such as maintaining open connections for extended processing times, preventing timeouts, and recovering from exceptions. The platform supports launching agent runs in the background, provides polling and streaming endpoints to monitor run status, and implements strategies to minimize exceptions. Features like heartbeat signals, configurable retries, and multiple streaming modes are crucial for reliable agent operation, providing end-users with intermediate output to demonstrate progress during lengthy processes.

A new paradigm called Group Think is being explored to further enhance the efficiency of multi-agent reasoning. This approach allows multiple reasoning agents within a single LLM to operate concurrently, observing each other's partial outputs at the token level. By enabling real-time mutual adaptation among agents mid-generation, Group Think reduces duplication and speeds up collaborative LLM inference. This contrasts with traditional sequential or independently parallel sampling techniques, which often introduce delays and limit the practicality of deploying multi-agent LLMs in time-sensitive or computationally constrained environments.

Recommended read:
References :
  • LangChain Blog: Why do I need LangGraph Platform for agent deployment?
  • AI News | VentureBeat: Beyond single-model AI: How architectural design drives reliable multi-agent orchestration
  • MarkTechPost: A Comprehensive Coding Guide to Crafting Advanced Round-Robin Multi-Agent Workflows with Microsoft AutoGen

Sean Endicott@windowscentral.com //
Microsoft is aggressively pursuing the integration of AI agents across its ecosystem, as highlighted at Build 2025. The company is embedding AI deeper into Windows 11, utilizing the Model Context Protocol (MCP) to facilitate secure interaction between AI agents and both applications and system tools. This move transforms Windows into an "agentic" platform where AI can automate tasks without direct human intervention. The MCP acts as a standardized communication layer, enabling diverse AI agents and applications to seamlessly share information and perform actions. Microsoft is also pushing AI to the edge with tools unveiled at Build 2025, and are creating smarter faster experiences across devices.

Microsoft is also enhancing its Microsoft 365 Copilot with "Model Tuning," allowing businesses to train the AI assistant on internal data, creating domain-specific expertise. This feature enables the creation of AI agents customized for specialized tasks, such as legal document creation or drafting arguments, using an organization’s unique knowledge base. It’s designed to secure data within the platform, ensuring that internal information isn't used to train broader foundation models. The feature is rolling out in June through the Microsoft Copilot Tuning Program, available to customers with at least 5,000 M365 Copilot licenses.

Adding to its AI advancements, Microsoft is exploring AI's role in various applications, like integrating Copilot into Notepad for AI-assisted writing and developing AI models like Aurora for accurate weather forecasting. However, a potential security concern arose when a private Teams message inadvertently revealed that "Microsoft is WAY ahead of Google with AI security" during a Build 2025 protest. The leaked message was within a Microsoft Teams message where Walmart are expanding their use of AI. The company is also developing NLWeb, an open-source protocol designed to AI-enable the web by transforming websites into AI-powered conversational interfaces.

Recommended read:
References :
  • Ken Yeung: Microsoft Adds Model Tuning to M365 Copilot to Build Domain-Specific AI
  • www.windowscentral.com: "Microsoft is WAY ahead of Google with AI security," says leaked message exposed accidentally following a protest at Build 2025
  • www.eweek.com: Microsoft’s Big Bet on AI Agents: Model Context Protocol in Windows 11
  • Ken Yeung: Microsoft Pushes AI to the Edge
  • eWEEK: Microsoft integrates the Model Context Protocol into Windows 11, paving the way for secure, AI-driven agents to interact with apps and system tools.
  • AIwire: Microsoft has introduced a new AI-powered orchestration system designed to streamline the complex process of cancer care planning.

@www.microsoft.com //
References: www.microsoft.com
Microsoft is introducing Magentic-UI, an open-source research prototype designed as a human-centered AI agent. This experimental tool is built to assist users in completing complex, web-based tasks in real time, directly within a web browser. Unlike fully autonomous systems, Magentic-UI emphasizes a transparent and controllable experience. The platform is geared towards tasks that are action-oriented and extend beyond simple web searches, providing a unique approach to human-AI collaboration on the web.

Magentic-UI builds upon Magentic-One and is powered by AutoGen, Microsoft's agent framework. It is available under the MIT license and on Azure AI Foundry Labs, offering developers, startups, and enterprises a space to explore Microsoft Research innovations. The system is integrated with Azure AI Foundry models and agents, with code samples available for those looking to integrate Azure AI agents into Magentic-UI's multi-agent architecture.

Magentic-UI is capable of tasks involving web browsing, Python and shell code execution, and file understanding. Key features include collaborative planning, where users can modify the agent's plan directly through a plan editor or textual feedback. It also supports collaborative execution, allowing users to pause the system, provide natural language feedback, or directly control the browser to guide the AI agent, fostering a seamless blend of human and artificial intelligence.

Recommended read:
References :

@blogs.microsoft.com //
Microsoft is aggressively pushing forward in the realm of Artificial Intelligence, as evidenced by several key initiatives and announcements. The company is focusing on the development and integration of AI agents, which are designed to enhance productivity and efficiency across various sectors. This commitment to AI innovation is underscored by the emphasis on building an open agentic web, facilitating collaboration and expansion within the AI community.

One significant aspect of Microsoft's AI strategy involves strengthening its position in software development. Visual Studio Code and GitHub are vital components of this strategy, used daily by millions of developers. To maintain its lead, Microsoft is planning to release the GitHub Copilot code extension inside Visual Studio Code under the open-source MIT license. In addition to integrating a new coding agent to GitHub Copilot at its Build conference. This approach is intended to encourage customization and contribution from the developer community, preventing fragmentation and solidifying Microsoft's standing in the coding world.

Beyond coding, Microsoft's Aurora AI foundation model exemplifies its capabilities in environmental forecasting. Aurora, developed by Microsoft Research, can predict a wide range of atmospheric events with greater precision, speed, and lower computational cost compared to traditional methods. Researchers have fine-tuned the model to predict ocean waves and tropical cyclones demonstrating its capability as a foundation model for the Earth system rather than just a foundation model for the atmosphere. Furthermore, Stanford is leveraging Microsoft’s agentic AI platform, Azure AI Foundry, to make advances in key healthcare areas.

Recommended read:
References :
  • The Microsoft Cloud Blog: Microsoft Build 2025: The age of AI agents and building the open agentic web
  • John Werner: Stanford’s Use Of Microsoft Agentic Platform Leads To Better Analysis

@www.eweek.com //
Microsoft is embracing the Model Context Protocol (MCP) as a core component of Windows 11, aiming to transform the operating system into an "agentic" platform. This integration will enable AI agents to interact seamlessly with applications, files, and services, streamlining tasks for users without requiring manual inputs. Announced at the Build 2025 developer conference, this move will allow AI agents to carry out tasks across apps and services.

MCP functions as a lightweight, open-source protocol that allows AI agents, apps, and services to share information and access tools securely. It standardizes communication, making it easier for different applications and agents to interact, whether they are local tools or online services. Windows 11 will enforce multiple security layers, including proxy-mediated communication and tool-level authorization.

Microsoft's commitment to AI agents also includes the NLWeb project, designed to transform websites into conversational interfaces. NLWeb enables users to interact directly with website content through natural language, without needing apps or plugins. Furthermore, the NLWeb project turns supported websites into MCP servers, allowing agents to discover and utilize the site’s content. GenAIScript has also been updated to enhance security of Model Context Protocol (MCP) tools, addressing vulnerabilities. Options for tools signature hashing and prompt injection detection via content scanners provide safeguards across tool definitions and outputs.

Recommended read:
References :
  • Ken Yeung: AI Agents Are Coming to Windows—Here’s How Microsoft Is Making It Happen
  • www.eweek.com: Microsoft’s Big Bet on AI Agents: Model Context Protocol in Windows 11
  • www.marktechpost.com: Critical Security Vulnerabilities in the Model Context Protocol (MCP): How Malicious Tools and Deceptive Contexts Exploit AI Agents
  • GenAIScript | Blog: MCP Tool Validation
  • Ken Yeung: Microsoft’s NLWeb Project Turns Websites into Conversational Interfaces for AI Agents
  • blogs.microsoft.com: Microsoft Build 2025: The age of AI agents and building the open agentic web
  • www.eweek.com: Microsoft’s Big Bet on AI Agents: Model Context Protocol in Windows 11