By Guest contributor |
Knowing where to get started with AI is a pressing concern for many IT and business leaders. The rapid evolution of the technology can be daunting, especially as challenges surrounding the responsible use of AI, privacy, security and hallucinations can further complicate implementation.
Even now, the public conversation around AI is shifting from generative to agentic. Looking forward, organisations must understand how they can leverage the technology across diverse use cases while maintaining a positive return on their investments.
The most effective approach to using AI involves enabling the connected experience via a framework designed to personalise user experiences, measure success through managed observability, and achieve higher levels of productivity through agentic AI solutions.
When implementing agentic AI, it’s beneficial to follow a progression of complexity which allows different use cases to be tested to ensure the business processes and technology are well understood. While this progression can vary between companies, the common route consists of five key levels.
Levels one and two focus on deploying AI-powered technology services and support in the back office, with use cases including self-help solutions, data access and email and calendar management.
Level three focuses on employee workflows, with key use cases in human resources such as automated onboarding and offboarding and tailored training materials. Then, level four focuses on the wider business processes and application modernisation through the automation of application refactoring aided by AI discovery tools.
Finally, level five focuses on consumer workflows and applies AI to the customer experience, with use cases including QR code-enabled virtual agents and personalised product suggestions.
A robust connected experience framework defines a set of priorities that IT leaders should focus as they progress through each level to implement agentic AI with the highest chance of success. For instance, personalisation is key for facilitating discovery and justifying investments, while creating experience level agreements are important for highlighting risks and guiding adjustments.
It’s currently estimated that between 70 and 85 per cent of AI projects fail to achieve their intended return on investment. While some of these failures can be attributed to the technical challenges presented by the rapid acceleration of AI, struggles can also be due to a lack of clear, measurable objectives or a weak data focus.
Security and the responsible use of AI are important factors for organisations to consider, and ensuring proper data handling is key. Furthermore, given that AI is designed to evolve over time, any tool being used must be properly governed to ensure no unexpected outcomes occur.
There are also serious cost considerations, especially as organisations are now spoiled for choice with the variety of AI tools available. Of course, not all AI is cost effective compared to the use case at hand, and solutions should be tailored to problems, not the other way around.
By focusing on delivering a connected experience, organisations can blend consultative strategies with traditional IT rollouts, emphasising personalisation and measurable outcomes. As adoption grows, industries will generate reusable agentic components, fostering innovation and confidence in AI-driven workflows.
With proper governance, technology selection and workflow optimisation, agentic AI can transform automation into a dynamic and impactful element of the enterprise ecosystem.
Dennis Perpetua is vice president and chief technology officer of Digital Workplace Services at Kyndryl
Visit the Kyndryl website to discover how it is helping organisations unlock the potential of AI
Discover more insights like this in the Summer 2025 issue of Technology Record. Don’t miss out – subscribe for free today and get future issues delivered straight to your inbox.