AI Agents and the Future of Data-Driven Innovation

AI

United Kingdom, Jun 3, 2025

An introduction to the transformative realm of Agentic AI

Authored by Scott Hodges, Head of Cloud, Logicalis UK&I, with contributions from from Igor Zanotti, AI Solution Architect at Kumulus (sister company of Logicalis within the Datatec Group).

Welcome to the first in a series of blogs that explores all things Data and AI. 

At Logicalis, we work directly with technology leaders like Microsoft and IBM which puts us at the forefront of cutting-edge solutions through platforms like Microsoft Fabric, Azure AI Foundry, Copilot Studio, IBM watsonx.ai, and watsonx.data.

In this first entry, we are excited to explore the rise of AI agents, the concept of Agentic AI, and introduce the Model Context Protocol (MCP) which, personally, I think is a game-changer that’s set to redefine how we collaborate and deliver value through technology.

We’ll also hear from Igor Zanotti, AI Solution Architect from Kumulus, who brings his deep expertise to the conversation, having recently presented at AWS all around MCP and Agents.

The Rise of AI Agents and Agentic AI

AI has evolved far beyond simple chatbots or predictive models. Today, we’re entering the era of Agentic AI. This is where AI systems don’t just respond to queries but act as intelligent, goal-oriented agents capable of reasoning, planning, and executing tasks with a degree of autonomy and self-orchestration. 

Think of AI agents as digital teammates that can: 

  • Handle complex workflows
  • Make decisions based on context
  • Integrate seamlessly with enterprise systems

An example that my team and I (shout out to Keith Madden and Seamus Flynn) worked on recently as an internal project, is a Sales Agent, that utilised a number of other domain-specific agents to fulfil its task as needed.

At Logicalis, we’re leveraging platforms like Microsoft Copilot Studio and Azure AI Foundry for customers to build AI agents that transform business operations. For example, Microsoft’s ecosystem allows us to create agents that can automate tasks like analysing data in Microsoft Fabric’s OneLake, or orchestrating multi-step processes. Likewise, with IBM WatsonX.ai and watsonx.data, we can develop agents that integrate with enterprise workflows, automating everything from IT (see AskIT), to onboarding, and supply chain optimisation. watsonx has some handy prebuilt agents out of the box to help accelerate the ROI which is a very important factor when deciding on AI use cases to focus.

AI agents are redefining how businesses operate by bridging the gap between data and action. With platforms like Azure AI Foundry, we're building agents that don't just process data but understand context and make decisions that align with business goals. This is particularly powerful in industries like finance and healthcare, where precision and speed are critical. What's exciting is that we're moving beyond isolated AI implementations to truly interconnected agent ecosystems. These agents can collaborate, share learnings, and coordinate complex multi-step processes that would traditionally require significant human oversight and manual handoffs between systems.

Ignor Zanotti
AI Solution Architect
Kumulus

What is the Model Context Protocol (MCP)?

One of the most exciting developments in the AI agent ecosystem to me is the Model Context Protocol (MCP), an open standard designed to enable seamless communication and context-sharing between AI agents and systems. The MCP is a framework (there are others such as A2A) that allows agents to exchange structured data and maintain consistent context across platforms.

Imagine a virtual assistant in a smart home system. The assistant needs to check the weather forecast from an online service, adjust the thermostat using a home automation app, and send a notification to your phone about the day’s schedule. Without a standardised protocol, these tasks could fail due to incompatible systems or data mismatches. A solution like MCP provides a common framework, allowing the assistant to seamlessly coordinate with these services, share information, and complete tasks efficiently.

MCP is a game-changer for multi-agent systems. At Kumulus, we're already seeing how it simplifies orchestration in complex environments, enabling agents to share context across platforms. This not only accelerates development but also ensures that agents deliver precise, data-driven outcomes. What makes MCP particularly compelling is that it was developed by Anthropic as an open standard, and we're witnessing its rapid adoption across the industry. All the major cloud providers—Microsoft, AWS, Google, and IBM—have adopted MCP, making it the de facto standard for agent communication. This standardisation is crucial because it eliminates vendor lock-in and enables true interoperability. We can now build agent architectures that seamlessly span multiple platforms and vendors, creating more resilient and flexible AI ecosystems for our customers. 

In fact, I recently presented a workshop about multi-agent systems and MCP at AWS for a major financial institution, where we were able to map numerous use cases for multi-agent systems, including customer quality assurance and risk assessment. The practical applications we identified demonstrated just how transformative these technologies can be when properly orchestrated through standardised protocols like MCP.

Ignor Zanotti
AI Solution Architect
Kumulus

Looking Ahead

The rise of AI agents and the adoption of standards like MCP are part of the rapid evolution of Data and AI. At Logicalis, what we’re focused on is working with customers to explore the use cases that are good candidates to run a proof of value on. Being able to rule use cases with little business impact out quickly, meaning we can spend more time focusing on one’s that deliver higher ROI. By focussing on higher value ROI use cases, customers realise their AI vision from use case, to real business value.

Our next entry will cover practical applications of AI in specific industries like finance, and manufacturing. 

Stay tuned!

 

Related Insights