By Joe Dunleavy, Global SVP, Head of AI Pod at Endava
In 2024, AI continues to dominate the news agenda, with its impact being felt in every sector.
As business leaders continue to realise the widespread benefits that can be gained from adopting AI-driven software, many are investing in a newer concept, agentic AI. Whilst still a relatively nascent technology, OpenAI recently made headlines with its prediction that agentic AI would become mainstream in 2025.
So, what are AI agents and why should businesses care?
First Things First
We know that AI can bring longer term value by taking on a greater degree of autonomy. Instead of relying on human direction for each separate task, agentic AI can advise on potential next steps and support data sharing with both large language models and cloud-based solutions.
This form of AI is often called multi-agent AI, as it works through empowering several ‘agents’. These agents have been assigned roles and are fed the necessary data, which enables them to complete tasks, communicate between themselves, and react to anomalies. Incredibly, they can also understand the context surrounding their actions so they can navigate any challenges and reach the best possible outcome while scaling at pace.
In public perception, agentic AI is a relatively new concept. Therefore, it’s understandable that business and consumers alike would be hesitant about feeding it private data without knowing exactly how it will be used. This apprehensiveness has been a challenge in the past, with organisations in highly regulated industries, like healthcare, reluctant to embrace the power of AI in order to preserve their user’s information and remain compliant.
Luckily, this is another area in which agentic AI can differ from traditional models.
A long-standing criticism of AI has been the lack of transparency around how the systems reach an outcome, with hallucinations and misinformation being a valid concern. With agentic AI, processes are visible and presented in a user-friendly format. The journey from data input to outcome is fully transparent, allowing those working in highly regulated industries to be able to show how and where private data is being used, which helps ensure it meets regulatory requirements.
While agentic AI empowers businesses to scale rapidly with greater efficiency, the necessity of human oversight remains paramount. In fact, agentic AI can be programmed to flag any queries to the people overseeing them, making the “human in the loop” an essential part of the process.
Next Steps for Businesses
For agentic AI to be rolled out successfully, businesses need a strong foundation. Addressing critical challenges such as data silos and outdated legacy systems is essential, as a weak infrastructure will hinder the full potential of any solution deployed.
As a practical example, think about chatbots. The person talking to the chatbot is looking for specific information about the business or product concerned. Without access to enough data on this topic, the chatbot will be unable to answer the question well, and the user will have to wait for a person to get back to them, causing unnecessary friction and making the entire process inefficient.
This analogy can be adopted on a wider scale when it comes to using AI agents. For the technology to enable large scale impact, it needs access to the right information from a data-based foundation. This is the fundamental first step.
Executives can then think about which form of AI technology is the best fit for them. For those keen to automate the smaller administrative tasks, like curating presentations and summarising documents, the most basic forms of generative AI will be satisfactory. Whilst useful for saving time and money, this tends to be a shorter-term fix than agentic AI, with more human intervention required to drive meaningful workflow forward.
Just The Beginning
We have only just scratched the surface of what agentic AI can achieve. While it offers a powerful edge through accelerated scaling and automation, its success is rooted in the strength of the underlying infrastructure. Once the base has been sufficiently modernised and powered by accurate data, businesses should feel confident when integrating agentic AI into their workflows. Whether working in a highly regulated industry, or wary of new technologies, business leaders are encouraged to feel safe in the knowledge that they can see the processes behind the outcome, making it easier to both avoid risks and ensure compliance.
