The integration challenge facing 95% of IT leaders with AI agents – and how to overcome it
Since the launch of Chat GPT, many business leaders have focused on what they thought was the right topic: Large Language Models ( LLMs). But these models are quickly becoming a commodity, as each one races to build the best for a specific use case.
Unlocking the full potential of AI
To truly unlock value from AI, you need to focus on everything around the model, such as the orchestration, the low code / no code approach to building and refining, the metadata framework and a data engine that compliments the data strategy. It's this platform advantage that is seeing agents across the globe stand up and deliver value with real data, leveraging real integration in a few short weeks.
To unlock the full potential of generative AI, a deeply integrated and connected platform with a single code base is essential. However, building such a platform requires significant time and resources unless you have already been empowering your employees with a comprehensive system.
By leveraging existing investments in digital infrastructure, businesses can seamlessly transition to an agentic workforce, enabling them to pivot with every flow, cloud, integration, and build—ultimately future-proofing their operations.
Agentic technology is a multi-trillion-dollar industry opportunity. The agentic enterprise will operate with unprecedented independence, capable of responding to queries and handling complex tasks autonomously. This autonomy will optimise workflows, drive innovation, and break down barriers related to the need for continuous human intervention.
By 2028, Gartner predicts that 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, allowing 15% of day-to-day work decisions to be made autonomously.
The role of data in AI agents
Yet, AI agents are only as good as the data they have. They need connected data—both structured and unstructured—to understand user queries and make informed decisions. That’s where integration and APIs come in, building a solid foundation for these agents.
While 93% of IT leaders are either implementing or planning to implement AI agents within the next two years, they face significant integration challenges that hold back the full potential of these agents.
According to the latest MuleSoft Connectivity Benchmark Report, which surveyed more than 1,000 IT leaders globally, 95% struggle with data integration across systems. On average, only 29% of applications are connected, which affects the accuracy and usefulness of AI agents.
The report found that, on average, enterprise organisations are using 897 applications, and those with AI agents are using even more -1,103 applications. 90% of IT leaders say data silos are creating business challenges.
The more applications and AI models there are, the harder it gets to integrate everything. Data silos make it even tougher, limiting agents' access to the data they need and leading to less accurate and useful outputs.
Disconnected data also places a major strain on IT resources. IT leaders are looking for ways to boost efficiency and productivity, but they expect their teams’ workloads to increase in the next year. Balancing current capabilities with integrating AI agents across hundreds of unique applications while maintaining those systems is a real challenge.
Simplifying integration for AI success
To unlock the full potential of AI agents, businesses need to align their integration and AI strategies. APIs and integration solutions can simplify and unify data infrastructure, allowing AI agents to access critical data and interact with existing systems and automation. This can significantly improve IT infrastructure, enable data sharing across teams, and integrate disparate systems.
Organisations that have successfully integrated their data and systems using APIs are reaping the rewards: increased productivity (49%), faster response to business needs (49%), and higher revenue generation (45%). On average, half of an organisation’s internal software assets and components are available for reuse, which means companies can leverage their existing investments instead of starting from scratch.
The reliance on IT teams highlights the need for a clear automation strategy, along with robust governance and monitoring to ensure everything runs smoothly and securely.
A well-rounded automation strategy is crucial for integrating AI effectively, but many teams are still working on theirs. One key part of this strategy is making AI accessible to non-technical users, which is essential for broader adoption and creating a solid foundation for employees to build on, and this is where agents are changing the game.
Every company, team, and employee will soon have an agent. But how useful is a team of agents if they can’t interact with other systems or agents to coordinate and take action across the entire business? AI must have a smooth handoff to a human, and if that transition isn’t well-coordinated and seamless, any benefits are quickly undone
As AI, integration, automation, and API use continue to drive transformation and performance, organisations that invest in these technologies to harness unlimited digital labour are best placed to stay agile, efficient, and ultimately succeed.
