Article
5 min read
Richard Pugh
  • SVP Global Head of Data & AI

Since the big data era, it feels like the hype around data and AI has been continually increasing.  In the last few years, with the advent of techniques such as large language models, the noise has reach unprecedented levels. 

 

This is driving leadership teams across all sectors to answer key questions around the way data and AI might represent an opportunity and a threat to their business.     

 

As we are a trusted digital transformation partner to some of the most recognised brands in the world, I spend a lot of time with leadership teams looking to understand how to chart a path forward against the backdrop of increasing hype, new innovations and pressure from internal and external stakeholders. During these conversations I often encounter an interesting perspective, usually accompanied by a resigned sigh – leaders will often say our data quality is poor so we’re a long way from getting value from AI’. Data quality is every organisation’s dirty secret. Their skeleton in the closet.    

 

“My data quality is poor” 

 

Firstly, let me just clarify that data quality isn’t a binary thing – data is never either ‘perfect’ or ‘awful’. There are many shades of grey, and every organisation has data quality challenges. 

When leaders tell me their data is poor, they are often referring to specific sets of data – perhaps customer data isn’t great and the CRM system has way too many names for every customer.  Perhaps the finance is split across many legacy systems and is held together by Sellotape and overly complex spreadsheets. 

 

But alongside this there will typically be strong data assets, perhaps driven by systems, by regulation or by the basic need for good data to oil the wheels of your business.I wanted to refute the aforementioned statement, to highlight the opportunities that can be unlocked with a set of impactful AI use cases that don’t rely solely on high-quality data assets. Let’s dive into those possibilities. 

 

Inspired by chess 

 

Before we look at the application of AI to create business value, I first want to talk about chess.   

We’ve been building mechanisms and algorithms that could play chess against humans for over 100 years. In 1997, IBM’s Deep Blue beat Garry Kasparov to become the first computer to beat a reigning world champion under standard tournament controls.Since then, increasingly sophisticated algorithms have been built that could pour over historical data on moves made to identify the optimal next move. The pick of the AI chess crowd was Stockfish, a free and open-source chess engine. 

 

In 2017, Google DeepMind’s ‘AlphaZero’ algorithm was developed, which beat Stockfish convincingly, heralding a new era of chess algorithm development.  AlphaZero was not trained on historical data – instead it used the rules of chess to play against itself thousands of times, teaching itself how to play.  In effect, it generated its own data through a test and learn strategy.  In total, the AlphaZero algorithm took only 4 hours of training to better Stockfish.So here we have an AI algorithm achieving amazing results without the need to train on historical data. 

 

AI use cases that don’t require high-quality data sources 

 

So which AI use cases can create significant business value without the need for high quality data assets?  Let’s start with 5 examples, but there are plenty more: 

 

1) AI enablement 

The first one is a little obvious, perhaps, but there are great ways to create tangible productivity gains across an organisation through the provision of co-piloting tools.  At Endava, for example, every employee has a license of OpenAI’s ChatGPT Enterprise, providing a rich set of functionalities backed by industry-leading language models.  We have seen productivity gains across a range of areas from software coding to sales presentation creation.  The key to this case is to engage the user community with the right training and support to enable adoption. 

 

2) Checks and balances 

We have had great success the automation of processes that are, essentially, a set of ‘checks and balances’.  Using agentic AI approaches, we can create autonomous agents who can perform a series of checks to support the automation of back-end process spanning everything from customer workflows to IT operations. These approaches tend not to involve historical data as they are more aligned to the chess example – a set of rules to be followed and calls to a set of APIs. 

 

3) Customer communications 

Another area where we have seen success is in the use of agentic AI to support basis customer communications. As an example, recently we worked on a project where an organisation routinely received information from companies that was incomplete, with people needing to send basic email correspondence back to those companies for clarifications or to request further information.  We were able to use AI to automate this basic customer interaction using agentic AI. 

 

4) Content creation 

We can use generative AI techniques to auto-create content based on specifications – without historical data, we rely on more detailed specification, or on set of rules and guidelines.  This could include the generation of marketing contents that conform to brand guidelines or initial drafts of business proposals that respond to specific sets of requirements.  We can also look to use AI agents that can reflect and check contents to ensure it is likely to be compliant to specific regulations or even resonate with a set of target personas. 

 

5) Insight extraction from documents 

We can use an agentic AI approach to read information from documents without necessarily needing rich historical data.  As an example, we recently delivered a project where we extracted rich company data from thousands of legal contracts to create a trusted data backbone for an organisation without the need for historical data. 

 

These are 5 examples of AI use cases that can create tangible business value (through productivity gains, cost reduction, increased customer satisfaction, compliance or direct revenue uplift) that can be invested in today. 

 

Strong data foundations remain the key to success  

 

So, you may be wondering whether this means we no longer need to worry about improving our data quality.  While we can create value through AI without data (as discussed above), the reality is that there are many AI use cases that require rich data. Beyond that, data is the lifeblood of a modern organisation and without high quality data a company is at a significant disadvantage. 

 

The reality is that if an organisation doesn’t build strong data foundations, this will impact their ability to react quickly as new AI capabilities emerge and mature.  This creates a significant risk as modern business continues to transform at pace in line with technological advancements. 

 

The good news is that there are many techniques that can help an organisation to build strong data foundations, but it is critical to start now for organisations looking to survive and thrive in a future business landscape. 

No video selected

Select a video type in the sidebar.