Cracking The Code: Why Interoperability Will Make Or Break AI Agents

The rise of artificial intelligence (AI) agents is reshaping how businesses operate, innovate, and compete on a scale not seen since the emergence of the internet. But before companies can unlock the true potential of the agentic enterprise, they need to crack the interoperability challenge. 

This is a key issue in the Middle East and globally, with 48 percent of information technology (IT) security leaders concerned that their data foundation isn’t set up to get the most out of agentic AI, while 55 percent aren’t fully confident they have the appropriate guardrails to deploy AI agents, according to a global survey of more than 2,000 enterprise IT security leaders.   

Capable of independent understanding, reasoning, and action, AI agents form a digital workforce that can manage high volumes of inquiries, automate routine processes, and offer real-time recommendations. Unlike simple chatbots, they learn from interactions, adapt to new situations, and maintain context in complex conversations, accelerating outcomes. 

AI agents are expected to boost productivity by 30 percent within the next two years as adoption jumps by 327 percent, according to Salesforce research. 

When people and AI collaborate, they achieve more than either could alone. Just as human employees rely on knowledge repositories and guidelines, AI agents need a flexible IT infrastructure to thrive. They must understand your business and access the latest information, working seamlessly with other agents across different technologies and vendors. 

Despite their impressive momentum, most agents still operate independently, handling specific tasks in isolation rather than collaborating effectively. 

The future of enterprise AI lies in seamless agent-to-agent collaboration. Achieving AI maturity and multi-agent orchestration across your entire digital landscape is the strategic imperative for organizations today to deliver unprecedented efficiency, productivity gains, and open up new business opportunities. 

Unlocking AI’s Potential, Step-By-Step 

A multi-agent ecosystem can orchestrate complex tasks across various applications and data sources, understanding nuanced differences in information, and empowering human workers to focus on higher-value activities. 

Take, for instance, a digital agent in customer service efficiently managing a product return request by seamlessly coordinating with a manufacturer’s agent for damage assessment and a logistics partner’s agent for shipping protocols, ensuring swift resolution. 

In sales, a digital sales agent can identify a new partnership opportunity, collaborate with a prospective partner’s sales agent to align on product offerings and pricing, and ultimately draft a tailored proposal and schedule meetings for human representatives. 

Without seamless connectivity to the hundreds of applications businesses rely on, agents are confined to narrow technical domains, limiting their ability to retrieve information, automate processes, and deliver meaningful outcomes. 

The journey to unlocking the full potential of AI agents is a step-by-step process, divided into four levels of sophistication. From helpful assistants (Level 1) and single-system task execution (Level 2) to handling end-to-end workflows across multiple systems (Level 3), the ultimate goal is Level 4: multi-agent orchestration. 

To advance through the levels, businesses should define clear objectives for their AI agent deployments and measure success. It's also crucial to harmonize data across various sources, ensuring quality and availability. Seamless integration with existing systems, focusing on user experience, is essential. 

A robust governance framework, including testing, monitoring, and security policies, is vital for auditability and transparent oversight. Additionally, planning for human oversight is key. 

Understanding Industry Protocols 

Achieving a collaborative agent-to-agent ecosystem goes beyond innovation and complexity. It requires common industry standards and partnerships between technology vendors.  

Without a common framework for agents to discover, authenticate, and communicate, the AI agent ecosystem becomes fragmented and siloed, missing the opportunity for richer, end-to-end automation.  

Open standards like the Model Context Protocol (MCP) and Agent-to-Agent (A2A) protocols are gaining significant traction and form the backbone for inter-agent communication. 

By supporting these standards, a trusted and open platform will enable secure connections between AI agents and diverse enterprise systems. This ensures agents communicate effectively, identify each other, and exchange information securely. 

However, relying solely on MCP and A2A can introduce critical challenges, particularly in governance, digital identity, and granular access control. Without enterprise controls, this could lead to data inconsistencies, security vulnerabilities, or compliance issues. 

Rethinking How Work Gets Done 

Transitioning to an agent-first enterprise isn't just about new technology; it’s about integrating intelligent, interconnected AI to break down silos and create a trusted hybrid workforce of humans and AI agents. This shift demands a re-evaluation of work processes to maximize efficiency and innovation. 

Agents capable of working with one another to complete tasks and projects can drive lasting business value only if they’re built on a solid digital labor platform, can access the right systems, apps, and data, and can find one another in industry-standard ways.  

This strategic approach sets the stage for a “Gen Agent” workforce, where AI agents serve as the new user interface, powering a projected $6 trillion digital labor economy, and ensuring business success in the dynamic AI era. 

About The Author 

Mohammed Alkhotani is the Senior Vice President and General Manager of Salesforce Middle East.