Newsletter #36: Anthropic's MCP and Google's A2A - The Leap in AI Agent Capabilities is Here, Now.

You’ve probably felt the tangible momentum around Enterprise AI Agents over the last 60 days and started to believe how real it is...and you’re absolutely right.

Last week’s Google Cloud Next conference continued to build on the momentum as they unpacked their vision and comprehensive AI Agent strategy. 

Included in the strategy was A2A -  “Agent2Agent (A2A), new open protocol that gives your agents a common language to collaborate no matter what framework or vendor they are built on.”

I’m going to try something a little bit different with Newsletter #36. 

Instead of hyping AI Agents in general as there is plenty of that already out there, I’m going to focus on two technical developments that are responsible for an enterprise AI leap in capabilities that are driving a whole lot of justifiable optimism. 

My hope is that by narrowing the focus a bit, this newsletter will provide readers with thoughts on two tangible developments to help move from AI Agent understanding to action.

The two developments are MCP by Anthropic and A2A by Google.

Anthropic's Model Context Protocol (MCP) provides AI Agents with a standardized, secure method to access and use tools, data sets etc. 

Think of MCP as a universal adapter or a USB port for AI.

Google's Agent2Agent (A2A) protocol establishes an open standard for different AI Agents, even those from competing vendors, to communicate and collaborate directly. 

Think of A2A as the next generation of API.

These two complimentary developments represent a leap in capability from Automation to Autonomy.

A leap that can be felt throughout conversations with AI-first enterprise leaders everywhere.

If Newsletter #36 doesn’t convince you to sprint in this direction, I haven’t done my job.

So let’s do this.

What is an AI Agent?

Let’s start by defining an AI Agent with a little help from McKinsey & Company.

Think of an AI Agent as an advanced software program that can understand and independently pursue goals. Instead of simply responding to specific commands, they can analyze situations, plan sequences of actions and adapt / iterate based on initial results, changing circumstances etc.

AI Agents represent a significant shift from reactive to proactive AI systems. 

Moving from basic automation characterized by predetermined actions and responses to autonomous goal achievement strategies, complex reasoning and problem-solving capabilities. 

Adaptive decision-making based on context, predictive analytics etc.

Similar to the way a human being aka “a knowledge worker” works...

They represent a significant leap beyond simpler AI tools such as basic chatbots or assistants. 

They can be thought of as digital coworkers as AI agents are designed to integrate directly into the flow of work, understanding context, interacting with various systems and taking actions, along with human coworkers, based on real-time signals.

Significance of MCP and A2A 

Now that we have the definition of an AI Agent out of the way, let’s dig into why MCP and A2A represent such a significant shift in enterprise AI capabilities.

Think of traditional AI systems as talented but isolated specialists, confined to their individual silos. They excel at specific tasks but struggle to access broader resources or collaborate with other systems. This limitation has been a critical bottleneck in realizing the full potential of enterprise AI.

MCP and A2A fundamentally solve this problem through complementary approaches:

MCP (Model Context Protocol) creates a universal standard for how AI agents connect to and leverage external resources—whether that's enterprise data, APIs, specialized tools, or legacy systems. 

It's like giving every AI agent a standardized set of adapters that work seamlessly with your organization's digital ecosystem. No more custom integrations or compatibility issues.

Equivalent of a USB port that represents a standardized way to plug and play.

A2A (Agent2Agent) takes this a step further by standardizing how AI agents communicate and collaborate with each other. 

Rather than having a single agent trying to handle complex workflows alone, A2A enables specialized agents to work together as coordinated teams - each bringing unique capabilities to solve parts of a larger challenge.

Together, these open protocols are creating a new paradigm of interoperability that marks a significant milestone in the maturation of the AI agent ecosystem. 

Rather than being locked into proprietary solutions from a single vendor, enterprises can now mix and match specialized agents from different providers to create powerful, customized solutions that address their specific business needs.

This convergence of technological capabilities - increasingly sophisticated AI models, standardized integration protocols, powerful development platforms, and proven ROI from early adopters - signals that we're approaching a tipping point. 

This is the tangible momentum around Enterprise AI Agents over the last 60 days we’ve all been feeling which is beyond exciting.

It isn't just the technology itself but the transformation it enables in how work gets done. 

These aren't futuristic concepts - they're emerging as genuine digital coworkers that will reshape enterprise operations from the ground up.

We're witnessing the birth of a new operating system for business - one where humans and machines collaborate in ways that allow each to focus on what they do best.

Needless to say but we’re still very early in the process and there is a ton that needs to get worked out in order to realize the potential of these proactive AI systems but there are clear tactical levers to consider pulling to keep your momentum going…

What are 3 critical levers that drive AI Agent + Human collaboration?

1: Context is King

Think of AI Agents and their on-boarding in the same way you would a new employee. 

When a new hire joins a company, the quality of their on-boarding, their understanding of leadership’s vision, strategy and tactical connection to resources, systems, data, resources etc are critical drivers of their ability to generate business outcomes.

The equivalent is true for AI Agents. 

The richer the context an agent possesses, the more relevant and accurate its decisions and actions will be. The ultimate effectiveness of enterprise AI agents is intrinsically linked to the quality, accessibility, and integration of an organization's underlying data and IT architecture. 

Just like there is no AI Strategy without a Data Strategy, the same can be said for enterprise AI Agents.

Sophisticated AI models cannot overcome the limitations imposed by siloed systems or poor-quality data. 

Therefore, organizations with well-managed data and integrated systems will possess a distinct advantage in leveraging these powerful AI Agents and intelligent systems.

2: Continuous Learning Generates Compound Returns

A big part of the enthusiasm around the shift from reactive to proactive AI systems is the simple and clear idea that AI Agents are built to do work.

While that is true and very exciting, it’s the knock on effect of AI Agents doing work where you move into the logic behind the upcoming era of abundance.

AI Agents are designed to be reflective, learn and improve over time based on their interactions, outcomes of their actions and associated feedback loops.

The more work AI Agents do, the more data exhaust the intelligent system gathers and is able to process into insights to make better decisions in the future.

The knowledge base is continuously added to, 24/7 by AI Agents, who autonomously figure out what the collective organization is learning, day-to-day so that everyone benefits from the continuous learning. 

Insights, ideas and direction that benefit all humans and all AI Agents.

Translating to previously unimaginable efficiency and productivity gains as an ensemble of humans and AI Agents collaborate over an outrageously valuable knowledge base.

One of the best descriptions of AI Agent and human collaboration I’ve heard in a long time was delivered by my man John Macintyre during a recent fireside chat we did. 

The quote can be found here at the 12:27 mark of our fireside chat as well on a LinkedIn carousel via slide 10 of the top 4 key takeaways of the talk.

3: 80/20 Rule + Cognitive Loads

When you look at the leap in capability enabled by MCP and A2A through Pareto’s Principle - 20% of your efforts generate 80% of your results - things get very exciting, very quickly.

One of the most immediate and tangible benefits of deploying AI agents is their ability to automate a wide range of repetitive, time-consuming tasks currently performed by human employees across various business functions.

For every 1% of time AI Agents give back to human collaborators that can be invested in the 20% that generates 80% of their results is big.

But the other part that can often be overlooked is the cognitive load that is also freed up and therefore the quality of that incremental 1%+ can not be underestimated.

This allows employees to focus on higher-value activities that require strategic thinking, complex problem-solving, creativity, and nuanced human interaction.

All of which are informed and enhanced by the intelligent system’s continuous learning referenced in #2 above aka augmented intelligence.

Within these systems, not only are humans able to focus more of their bandwidth on higher impact work but they’re also enabled or augmented accordingly. 

The quality of the augmentation is where you can start to see the promise of each knowledge worker effectively being provided with 5-10x the firepower they had before without the leverage of AI Agents and intelligent systems.

AI Agents Don’t Just Optimize Existing Processes, They Enable Entirely new Capabilities That Were Previously Impossible Due to Resource Constraints or Technical Barriers

Traditional enterprise software has always been constrained by integration challenges and rigid workflows. The true promise of AI Agents is revealed by technical breakthroughs such as MCP and A2A that remove those limitations.

Standardizing how AI Agents connect to tools and other agents to unlock the promise of multi-agent systems enable companies to unleash specialized intelligence exactly where it’s needed most, 24/7.

This is a race that everyone has an opportunity to run as the starting line has suddenly become crystal clear. The choice of when and how to get started is yours.

Gong's 14 AI Agents

While we've explored the theoretical potential of MCP and A2A above, pioneering SaaS companies have been aggressively investing in their enterprise AI enhanced features over the last 12+ months to pull the future forward.

One of those companies is Gong, the revenue intelligence company that recently launched 14 AI Agents.

Last week, I read this excellent LinkedIn post by Udi Ledergor, Gong's Chief Evangelist. 

In it, he stack-ranked his top 3 of the 14 agents Gong launched. 

This isn't a proof-of-concept or a limited pilot - it's a comprehensive deployment that showcases exactly how the standardized connections enable specialized agents to work together in solving real business challenges.

While the post speaks for itself, I couldn’t agree more with Udi’s ranking of AI Briefer as #1.

Competing on customer experience is one of the last and only sustainable competitive advantages in business.

Truth be told, it’s very hard to do because it requires a meticulous attention to detail that can be quite tedious, but critical to maintain.

This is where all of the above can come to life.

Imagine how much you can improve your CX quotient when you deploy a multi-agent team to collaborate with human, customer facing teams by enabling each of them to have a complete understanding of everything / anything that’s relevant, up to the minute?

Imagine how much more effective your rockstars that love solving your clients' most difficult problems can be when they can outsource so much of the tedious to your AI Agent team?

Imagine how much more strategic, creative and empathetic your client-facing team will be when you give them back all of that cognitive load so they can think deeply and executive flawlessly for your clients?

You see what I mean?

Enterprise AI Agents. For the win.