AI Agents Are Learning to Talk

Happy Monday!

In the fast-evolving world of AI, two groundbreaking protocols are setting the stage for a new era of intelligent collaboration: Google's Agent2Agent (A2A) and Anthropic's Model Context Protocol (MCP). These protocols aim to standardize how AI agents communicate and interact, potentially transforming industries and redefining human-AI interaction.

🧠 What Is Agent2Agent (A2A)?

Launched on April 9, 2025, by Google, the Agent2Agent (A2A) protocol is an open standard designed to facilitate seamless communication between AI agents across different platforms and vendors. With support from over 50 technology partners—including Atlassian, Salesforce, and PayPal—A2A enables agents to exchange information and coordinate actions securely, enhancing interoperability in complex enterprise environments.

Key Features of A2A:

• Agent Discovery: Agents can advertise their capabilities using standardized "agent cards," allowing for dynamic discovery and collaboration
• Flexible Communication: Supports various communication methods, including HTTP, Server-Sent Events (SSE), and push notifications, accommodating both short and long-running tasks
• Security: Built with enterprise-grade authentication and authorization, ensuring secure interactions between agents
• Modality Agnostic: Designed to handle various data types, including text, audio, and video, enabling rich, multi-modal interactions

🔌 What Is the Model Context Protocol (MCP)?

Introduced by Anthropic in late 2024, the Model Context Protocol (MCP) provides a standardized way for AI models to access external data sources and tools. By simplifying the integration process, MCP allows developers to connect AI assistants to various datasets and applications without extensive custom coding.

Key Features of MCP:

• Standardized Context Delivery: Enables consistent and secure provision of contextual information to the AI model.
• Plug-and-Play Integrations: Facilitates easy connections to multiple data sources, reducing development time and complexity.
• Vendor Flexibility: Designed to work across different AI systems, minimizing vendor lock-in and promoting interoperability.  

🔄 Complementary Roles: A2A and MCP

While A2A and MCP serve different purposes, they are complementary in building robust AI ecosystems.

• *MCP focuses on connecting individual AI agents to external data and tools, providing the context for informed decision-making.
• *A2A enables multiple AI agents to communicate and collaborate, coordinating actions across various systems and platforms.

Together, they facilitate the development of intelligent, interconnected AI systems capable of complex, multi-agent workflow.

🌐 A Glimpse into the Future

Integrating A2A and MCP protocols heralds a future where AI agents can seamlessly interact, share information, and coordinate actions across diverse platforms. This evolution promises to enhance efficiency, foster innovation, and unlock new possibilities in AI-driven solutions.

As these protocols gain traction, we can anticipate a surge in intelligent, collaborative AI systems that will transform industries and redefine the boundaries of automation.

And now, here is the AI news from last week.

Not a subscriber yet? Click here to subscribe.

Notable AI news from Last Week

The $40 Billion Startup Mystery Shaking Up Silicon Valley:  Figure AI, a three-year-old robotics startup founded by serial entrepreneur Brett Adcock, is making headlines with its audacious attempt to raise $1.5 billion at a jaw-dropping $39.5 billion valuation—despite having no revenue and just a handful of robots in deployment. The company has sparked excitement and skepticism in Silicon Valley by promising to revolutionize industrial and domestic automation with humanoid robots, aiming to deploy over 200,000 units by 2029. 

Google Announced Ironwood TPU in Las Vegas: On April 9, 2025, at the Cloud Next 2025 conference in Las Vegas, Google unveiled its seventh-generation Tensor Processing Unit (TPU), named Ironwood. This TPU is specifically engineered to handle the demands of inference workloads in generative AI applications, such as large language models (LLMs) and advanced reasoning tasks.  Although GPUs have consistently outperformed AI inference-developed specific chips like TPUs to date, Google hopes that Ironwood will finally prove that an AI-specific chip can outperform a general purpose GPU.    

Andreessen Horowitz seeks to raise $20 billion megafund amid global interest in US AI startups: According to Reuters, Andreessen Horowitz is aiming to raise approximately $20 billion—the largest fund in its history—to tap into strong global investor demand for US-based AI companies. As communicated to its limited partners, the fund will reportedly focus on growth-stage investments.

The AI Race Has Gotten Crowded—and China Is Closing In on the US:  The 2025 AI Index from Stanford University's Institute for Human-Centered AI highlights the rapid diversification and globalization of artificial intelligence development, marking the end of the previous dominance by just two American giants—OpenAI and Google. The emergence of China's DeepSeek-R1 as a top-tier model, despite US efforts to limit China's technological resources, underscores this shift. This accelerating global competition points to a more diverse and unpredictable future landscape in AI, pushing the boundaries closer toward achieving artificial general intelligence (AGI).  

Dr. Oz Pushed for AI Health Care in First Medicare Agency Town Hall: Dr. Mehmet Oz, newly appointed administrator for the Centers for Medicare and Medicaid Services (CMS), is making waves by prioritizing artificial intelligence in patient care, notably suggesting that AI avatars could potentially replace frontline healthcare workers, significantly reducing patient costs. While CMS has previously explored AI to analyze large datasets, Oz's proposal signals a controversial shift toward direct AI involvement in clinical practice, despite skepticism from medical research and current CMS staff. This bold push highlights ongoing tensions and debates over AI's appropriate role and limitations within critical healthcare policy and patient care environments.   

Musk's DOGE using AI to snoop on U.S. federal workers, sources say:  According to a recent Reuters report, Elon Musk's Department of Government Efficiency (DOGE) is utilizing artificial intelligence to monitor U.S. federal employees' communications for perceived disloyalty to President Donald Trump and his administration. This surveillance includes scanning internal communication platforms like Microsoft Teams for anti-Trump or anti-Musk sentiments. Encrypted messaging apps, such as Signal, which allows messages to disappear, have raised transparency and legal concerns, potentially violating federal record-keeping requirements.    

The AI Agent Era Requires a New Kind of Game Theory:  Zico Kolter, a professor at Carnegie Mellon University and OpenAI board member, emphasizes the need for a new form of game theory to address the security challenges of autonomous AI agents. As AI systems become more capable of independent interactions, they introduce complex dynamics that traditional models fail to capture. Kolter's work focuses on developing inherently secure AI models and understanding the strategic behaviors that emerge when AI agents interact. This perspective is crucial for ensuring that as AI agents evolve, their interactions remain safe and predictable, highlighting the importance of integrating advanced game-theoretic approaches into AI development.    

Fintech founder charged with fraud after ‘AI’ shopping app found to be powered by humans in the Philippines:  Albert Saniger, founder and former CEO of Nate, a once-hyped AI-powered shopping app, has been charged with defrauding investors. Despite raising over $50 million from major VC firms like Coatue and Forerunner Ventures, the DOJ alleges that Nate's claims of “AI-driven universal checkout” were a façade — the app was powered almost entirely by manual labor in the Philippines. Nate’s pitch hinged on an AI that enabled one-click purchases across any e-commerce site, but according to prosecutors, its automation rate was virtually zero.    

Announcing the Agent2Agent (A2A) Protocol:  Google has just announced A2A (Agent-to-Agent), a groundbreaking open standard designed to usher in a new era of interoperability between AI agents 🤖↔️🤖. This initiative aims to make it easier for different AI agents—from various companies or ecosystems—to discover, communicate, and collaborate securely. Backed by major players like NVIDIA, Hugging Face, LangChain, and UC Berkeley, A2A is poised to accelerate the development of complex, multi-agent systems by standardizing how agents interact. This matters because it could shift AI from siloed tools to a cooperative agent economy, where AI systems from different platforms work together—dramatically boosting productivity, automation, and innovation. 🌐🚀    

Shopify CEO says before hiring anyone new, employees must prove AI can't do the job better:  Shopify CEO Tobi Lütke has mandated that before hiring new employees, managers must demonstrate that artificial intelligence cannot perform the job more effectively. In an internal memo titled "AI usage is now a baseline expectation," Lütke emphasized that proficient use of AI is a fundamental skill for all Shopify employees, including executives, and will be evaluated in performance reviews. This initiative follows significant organizational changes, including layoffs and acquisitions of AI-focused startups, reflecting Shopify's commitment to integrating AI to enhance efficiency and innovation.

Is SEO is dead? Long live AIO!:  The rise of AI tools like ChatGPT, Claude, and Gemini is rapidly changing how people discover information—especially in B2B. Instead of relying on traditional search engines and SEO strategies, users ask AI assistants directly for insights, comparisons, and recommendations. This shift from search engine optimization (SEO) to large language model optimization (AIO) has enormous implications for marketers. Companies must now tailor content for machines, not just humans, as AI becomes the primary gateway for discovery and decision-making.

OpenAI countersues Elon Musk, claims harassment:  OpenAI has filed a countersuit against Elon Musk, accusing him of harassment and asking a federal judge to prevent him from taking any further "unlawful and unfair actions" in the ongoing legal dispute over OpenAI’s transition to a for-profit structure.

Notable Podcasts and Videos

NPR Why AI is entering your kitchens and living rooms

NPR Research finds how AI will impact demographics differently

Recent Investment Activity

Tracking 32 AI companies last week that raised $2.82B, here are the highlights:

Anecdotes – A Palo Alto startup automating governance, risk, and compliance processes with AI. Raised $30M in a Series B extension.

Arena – Based in New York, this company helps hardware engineers optimize electronic devices through AI. Closed a $30M Series B.

Artisan – A San Francisco startup building AI-powered digital employees, starting with sales automation. Raised $25M Series A.

Aurascape –A Santa Clara-based firm protecting enterprise AI applications with a native security platform. Raised $50M.

BLNG AI – A Los Angeles company offering AI tools for transforming jewelry sketches into photorealistic designs. Raised $3M seed round. 

Bliss Aesthetics – This Encinitas startup connects patients with plastic surgeons and uses AI to visualize outcomes. Raised $17.5M.

Blue Water Autonomy – Boston-based startup developing uncrewed, autonomous naval ships. Raised $14M seed round.

Conduit – A San Francisco company automating workflows for property managers across housing and hospitality. Raised $3.1M.

DIAMO – New York startup using AI for hotel revenue optimization, digital marketing, and booking. Raised $4M seed round.

Gallatin – Washington, D.C.-based company enhancing military logistics with AI. Raised $15M seed round.

HoneyHive – New York startup building tools to evaluate and improve the performance of AI agents. Raised $5.5M.

illumicell AI – Boston startup delivering real-time AI diagnostics for male fertility testing. Raised $2M pre-seed.

Incident.io – Headquartered in New York and London, it streamlines engineering incident response with AI. Raised $62M Series B.

Inventex – Salt Lake City company using AI agents to identify inventions and generate patent filings. Raised $2.4M pre-seed SAFE. 

Krea – San Francisco platform that unifies generative AI models for creative image workflows. Raised $47M Series B.

LiveKit – San Jose-based open-source platform for building low-latency AI agents with media capabilities. Raised $45M Series B.

nEye Systems – Berkeley firm developing optical switches to improve speed and energy use in data centers. Raised $58M Series B.

NexGen Cloud – London startup offering cloud-based access to high-performance GPUs for AI computing. Raised $45M.

Nuro – Mountain View-based autonomous vehicle company now expanding into robotaxis. Raised $106M Series E, with $2.2B raised to date.

Octane – San Francisco startup analyzing blockchain smart contracts for security flaws using AI. Raised $6.75M.

Outtake – Brooklyn startup using AI agents to protect organizations from impersonation and scams online. Raised $16.5M Series A.

Parallel Systems – Los Angeles company developing autonomous electric rail vehicles. Raised $38M Series B.

Qevlar AI – Paris-based firm using AI agents to investigate and respond to cyber threats. Raised $10M.

Rondah AI – A New York startup bringing virtual AI receptionists to dental practices. Raised $1.8M pre-seed.

Sagittal AI – London startup building AI coding assistants that review and update software documentation. Raised $2.2M pre-seed.

SandboxAQ – Palo Alto startup combining AI and quantum software for cybersecurity and drug discovery. Raised $150M Series E extension. 

Series – A New York-based AI-powered social network tailored for college students. Raised $3M.

SigIQ.ai – Berkeley startup offering personalized AI tutors for test prep. Raised $9.5M seed round.

Solve Intelligence – A London company with an AI document editor designed for patent attorneys. Raised $12M Series A.

Sourgum – Jersey City firm streamlining commercial waste and recycling with AI. Raised $12.5M Series A.

Starhive – Stockholm-based startup optimizing IT workflows and resources using predictive AI. Raised $5M.

Thinking Machines Lab – A new AI venture founded by ex-OpenAI CTO Mira Murati. Aiming to raise $2B at a $10B+ valuation.

YRIKKA – New York company building red teaming tools for cybersecurity using AI. Raised $1.5M pre-seed.

Thank you

Thank you to Bob Stefanski for contributing articles during the week.

That's all from last week,

Doug Neal

#eLabAIReport #ai #artificialintelligence #deeptech #AIreport #A2A #MCP #AIAgentsTalking eLab Ventures