Harnessing agentic AI

Joe Dunleavy at Endava shares some practical steps to success

 

Agentic AI has the potential to transform industries. However, according to Forrester, three quarters of the firms who build their own aspirational agentic architectures will fail. To keep up with the competition, business leaders must consider how they can best deploy the technology and make meaningful, strategic investments in innovation. 

 

 

What is agentic AI?

Agentic AI helps organisations scale and simplify their daily operations by automating complex workflows. This advanced technology takes the forms of agents, each given personas and decision-making capabilities. As opposed to undertaking tasks based on prompts like Gen AI, agentic AI helps businesses to automate more complex workflows, interpreting nuanced contexts to reach the objectives it has been set. 

 

This can reshape workflows and enable human employees to focus on high-value areas where their expertise makes the greatest impact. By analysing data in real time and making decisions accordingly, agentic AI can fundamentally transform a corporation’s ability to be flexible in ever-changing market conditions.

 

Forrester’s findings highlight the risks of failure, but organisations can succeed by taking the right steps to prepare. 

 

 

How to ensure success

For businesses, adopting agentic AI is a big decision and, like most innovations, it requires smart planning to succeed. In a recent survey, only 37% of respondents believed that their organisation is adequately prepared to implement AI. 

 

Many companies are navigating decades of legacy technology, which can understandably feel overwhelming, especially if they’re unconvinced about embracing AI in the first place. Data quality is complex, and if you feel you need perfect data before getting started, you might find yourself stuck on the starting blocks.

 

But data quality isn’t binary – data is never perfect or awful. There are many data sets throughout any company, and some of them will be stronger and cleaner than others.

 

So my first advice is to start small. Identify high-impact use cases that have the potential to bring significant return on investment, and get alignment across the organisation to ensure these are the projects that will deliver the greatest value.

 

Consider engaging with external industry experts here. Whilst you know your business better than anyone else, external partners can shed light on how others in your industry operate, and provide an alternative perspective which can be incredibly useful when implementing new technologies. 

 

Further along the line, you can revisit the challenge of improving your data quality. The reality is that there are many AI use cases that require rich data and beyond that, data is the lifeblood of a modern organisation. By building stronger data foundations in the long term, businesses will greatly improve their ability to react quickly as new AI capabilities emerge and mature. 

 

 

Highly regulated industries

Agentic AI is gaining traction in highly regulated industries such as banking, insurance and healthcare – and for good reason. Technology innovation holds immense potential to drive positive change, such as streamlining healthcare processes to deliver quicker treatments, or make faster decisions around processing insurance claims.

 

But despite the many benefits, there is natural concern about the steps taken to reach its goals. Even with full confidence that a tool is complying with tough regulatory standards, it’s understandable for those working with it day-to-day to still pose the question, ‘how does it work?’.  

 

Unlike many other forms of AI, which operate in something of a ‘black box’, agentic AI models have the potential to give full transparency into what they are doing. Firstly, agents must be given ironclad “data contracts” – a highly specific set of instructions which are not open to interpretation. Once this data flow is confirmed, AI can be introduced to action it. 

 

Then, when underpinned by an industry standard data foundation with lineage and auditability built in, the data being used - and the decisions made based on it - are logged and shared in a way that is easy for users to understand. This allows employees to work alongside agentic AI as colleagues, trusting that tasks are being carried out appropriately, because they can see how information is being handled and decisions are being made.

 

This is again an area where external support can be helpful, both for understanding the workflows where agentic agents can add the most value, and for training employees to work alongside them.

 

 

Next steps for leaders

Agentic AI has the potential to bring high levels of operational efficiency to complex workflows, enabling the people within your business to focus on value-driven activities like customer interaction and problem solving.

 

But balancing innovation with compliance and skills can be a difficult task for any organisation, and the implementation of agentic AI is no different. Through consulting experts with specific area expertise, leaders can move forward with confidence, implementing AI to meet both industry expectations and corporate goals.

 

By taking a thoughtful, expert-led approach, organisations can unlock the value of agentic AI and be better positioned for success in 2025 and beyond.

 

---

 

Joe Dunleavy is EMEA CTO at Endava

 

Main image courtesy of iStockPhoto.com and PhonlamaiPhoto