Investment Opportunities in The Protocol Layer of an Agentic AI Future
Agentic AI: A Paradigm Shift in Computing
An agentic AI future envisions a world where intelligent, autonomous agents act on our behalf, performing complex tasks, and collaborating with each other to achieve goals. This is a significant leap from the previous generation of AI, which is largely comprised of passive tools that require direct human instruction for every action.
Today's AI vs. The Agentic Future
Today, we use AI as a tool. We ask a question and get an answer. We give a command and it's executed. The user is the "agent" orchestrating these tools. In an agentic future, AI agents are proactive and autonomous. For example, a team of specialized agents could collaborate to launch a marketing campaign, from analyzing market data to generating ad copy, creating visuals, and deploying the campaign across multiple platforms.
This future promises a world of increased productivity, efficiency, and creativity. By offloading complex, multi-step tasks to a network of capable agents, we can focus on higher-level strategic thinking and problem framing.
To realize this future, several key components are necessary:
Advanced AI Models: Foundational models with sophisticated reasoning, planning, and tool-use capabilities.
Specialized Agents: A diverse ecosystem of agents with specific skills and expertise (e.g., a travel agent, a coding agent, a financial analysis agent).
A Common Language: Standardized protocols that allow these disparate agents, built by different developers on different platforms, to communicate and collaborate seamlessly.
Robust Infrastructure: A new stack of technologies for agent discovery, security, identity management, and orchestration.
The Role of AI Agent Protocols
AI agent protocols are the bedrock of a truly interoperable agentic ecosystem. They are the standards and rules that govern how agents communicate, share information, coordinate their actions, and interact with external services. It's important to distinguish between AI agent protocols and orchestration.
Agent Orchestration refers to the management and coordination of multiple AI agents within a single, often proprietary, system to accomplish a complex task. An orchestrator, typically a central AI agent or framework, directs the workflow and assigns tasks to specialized agents.
AI Agent Protocols are open standards that enable communication and collaboration between agents across different, independent systems. They are the "lingua franca" that will allow a Salesforce agent to talk to a Google agent, or an open-source agent to interact with a corporate one.
Without open protocols, we risk a future of walled gardens, where a handful of large tech companies control their own agentic ecosystems, limiting innovation and user choice. Protocols are the key to unlocking a truly open and competitive agentic landscape.
MCP and A2A: The Foundational Protocols
Two emerging protocols are laying the groundwork for this open ecosystem: the Model-Context Protocol (MCP) and the Agent-to-Agent (A2A) Protocol.
Model-Context Protocol (MCP): Developed by Anthropic, MCP standardizes how an AI model connects to external tools and data sources. Think of it as providing a universal interface for agents to access databases, APIs, file systems, and other tools. This "vertical integration" allows a single agent to be more capable and context-aware.
Agent-to-Agent (A2A) Protocol: Spearheaded by Google, A2A is an open standard that enables communication and collaboration between autonomous agents. It allows one agent to discover the capabilities of another, assign tasks, and exchange information. This "horizontal integration" is crucial for multi-agent collaboration.
MCP and A2A are typically seen as complementary protocols. An agent might use MCP to access a specific tool (e.g., a flight booking API), and then use A2A to communicate its findings to another agent (e.g., a calendar management agent) to coordinate the next step in a workflow.
Beyond MCP and A2A: The Next Wave of Protocols
While MCP and A2A are foundational, a mature agentic ecosystem will require a richer set of protocols, including:
Agent Identity and Authentication Protocols: A standardized way for agents to prove their identity and access permissions. This is crucial for security and for enabling agents to act on a user's behalf. The evolution of standards like OAuth for agents will be critical here.
Value Exchange and Micropayment Protocols: As agents perform tasks for each other, they will need a way to exchange value. Protocols for seamless, low-friction micropayments will be essential for a thriving agent economy. This could leverage existing blockchain infrastructure.
Reputation and Trust Protocols: A decentralized system for establishing the reputation and trustworthiness of agents will be necessary to prevent malicious or low-quality agents from proliferating.
Data Provenance and Verifiability Protocols: To ensure the reliability of information shared between agents, we will need protocols for tracking the origin and history of data.
The New Infrastructure and Application Stack
The rise of agent protocols will catalyze the development of a new wave of infrastructure and applications. Some examples include:
Infrastructure:
Agent Registries and Discovery Services: "App stores" for agents, where users and other agents can find and assess the capabilities of available agents. At this point, we’re probably still thinking about this problem too literally, and a more appropriate framework for agent discovery than the store or registry model will likely emerge.
Agent Security and Monitoring Platforms: Tools for securing agent-to-agent communication, monitoring agent behavior, and preventing malicious activity.
Decentralized Identity and Access Management (IAM) for Agents: Solutions that allow users to manage the permissions and capabilities of their agents in a secure and granular way.
Agent-Native Data Storage and Management: Databases and data infrastructure specifically designed for the needs of autonomous agents.
A way to think about the opportunity here is, “What are the tools, systems, and infrastructural services needed on top of agent protocols to enable agents to work together (or individually) and accomplish their goals seamlessly?” And this would include not only infrastructure for users and developers of agents, but also infrastructure that agents would access and utilise directly.
Applications:
Multi-Agent Orchestration Platforms: While distinct from protocols, these platforms will leverage open protocols to allow developers to build and manage complex, multi-agent workflows across different ecosystems.
"Agent-as-a-Service" Platforms: Enabling developers to easily create, train, and deploy specialized agents that are discoverable and interoperable via standard protocols.
Consumer-Facing Agent Management Applications: "Mission Control" for your personal AI agents, allowing you to manage your team of agents, set goals, and monitor their progress.
Enterprise Agent Marketplaces: Platforms for businesses to discover, procure, and deploy specialized agents for various business functions.
Here, we’re potentially looking at applications and platforms that enable users and developers of agents to discover, deploy and manage agents for both consumer and enterprise use cases.
Investment Approaches in the Agentic AI Landscape
The agent protocol space represents a greenfield opportunity for investment. While the protocols themselves are likely to be open standards, the infrastructure and applications built on top of them offer significant potential for value creation. Potential investment areas include:
The Infrastructure for Agent Autonomy:
Agent Security: Companies building the firewalls, identity solutions, and monitoring tools for the agentic economy.
Agent-Native Infrastructure: Startups developing the databases, hosting platforms, and development tools specifically for building and deploying autonomous agents.
Agent Discovery and Registries: The "Google" for AI agents, enabling discoverability and trust.
The New Application Layer:
Cross-Platform Orchestration: Platforms that allow businesses to build and manage agentic workflows that span multiple ecosystems (e.g., Salesforce, Google, and open-source agents all working together).
Specialized "Agent Foundries": Companies focused on building and deploying highly capable, specialized agents for high-value industries like commerce, finance, healthcare, and law.
Enterprise Agent Management and Governance: Tools that allow large organizations to manage their fleet of AI agents, ensure compliance, and track ROI.
The Emerging Agent Economy:
Agent-to-Agent Payment Rails: Platforms that facilitate seamless value exchange between agents, potentially leveraging blockchain technologies.
Decentralized Reputation Systems: Building the "trust layer" for the agentic internet.
A Foundational Investment for the Next Computing Paradigm
In the context of an agentic AI future, we believe that the protocol layer, and the infrastructure and applications that will be built on it, represent a foundational investment opportunity on par with the early days of the Internet.
The development and adoption of open protocols like MCP and A2A is a critical inflection point. It signals a move away from a fragmented landscape of siloed AI tools towards a truly interoperable and collaborative agentic ecosystem. For investors, this is the time to invest in the companies that are building the foundational infrastructure and the breakout applications that will define this new paradigm, while the race to build the future of agentic AI is just beginning.
If you liked this article and are reading it on the web or received it from a friend, please consider subscribing to my regular newsletter (so you’ll get articles like this delivered fresh to your inbox) by clicking the subscribe button below.