Fujitsu: how to have a proper conversation with your data
Nick McDonald, Senior Account Director at Fujitsu, says ChatGPT has been a revolution, spawning the idea for a product that can help operators and suppliers really understand the data they generate and hold.
Data is a challenging area for operators and suppliers, and it’s easy to see why. First, there is the sheer volume of data generated, from KYC and customer onboarding to payments, bonuses, player behaviour and monitoring, compliance and legal.
This data needs to be collected, cleaned, stored, analysed and actioned, often in real-time. That’s because data is highly valuable, especially when it comes to making business-critical decisions. But for this, operators need to be able to extract the insights the data holds, and this can be tough given the data infrastructure and processes at most organisations.
Data is usually siloed and hard to access for certain members of the team, and it often needs cleaning and translating into a consistent format so it can be properly analysed and actioned against – which very rarely happens. For many, they have the power of data in their hands but don’t have the infrastructure or indeed the tools or knowledge to use it.
ChatGPT has changed how we engage with data
In a personal capacity, many of you reading this will have used or be a regular user of ChatGPT which allows users to ask questions of data and be presented with answers in an instant. The rise of ChatGPT has been nothing short of meteoric, taking just two months to reach 100 million users – it took Facebook 54 months to reach the same milestone.
ChatGPT uses the phenomenal power of Generative AI to extract value (information, in this case) from a vast array of data sources and present it in a way that is easy to digest and understand. For most people, ChatGPT is just a bit of fun and a good tool for learning more about a particular topic or perhaps gathering data and information for a report.
But could you imagine the potential of applying its capabilities at an organisational level, and having it ask questions of the company’s proprietary data? This would be an incredibly powerful tool that would allow operators – and suppliers – to extract the full value of, and the insights held within, the data they generate.
This is not possible with ChatGPT because it runs on publicly-available data and also consumes the information that is inputted into it – so organisations would be making their data public. There are also concerns about the reliability of ChatGPT which would make it hard to use for certain tasks, especially things like searching for legal precedence or compliance requirements.
Introducing the concept of a Private GPT
The important part of ChatGPT is the GPT bit which stands for Generative Pre Transformer and is the thing that trawls through and analyses data, and then generates the response to the query. With ChatGPT, this is public but for an organisation to use it on its own data, it must be private so that it can feed off the proprietary and sensitive information generated by the business.
If this is private, employees can then ask questions of the proprietary data without it being consumed into the public domain. Just think how revolutionary that would be. Instead of organisations struggling to really understand the data they hold, employees can literally ask it questions and generate answers in seconds.
The use cases for a Private GPT go way beyond asking questions of the entire data set and include greater collaboration between teams with the opportunity to jointly generate insights. They can also help evaluate results and decision support systems, and even roll out innovative consumer-facing products.
Let me explain by way of example. An online sportsbook operator could use a Private GPT to support its legal and compliance teams when entering new markets. The teams can ask questions and have the Private GPT trawl through legal and compliance documents in all markets where the operator is active and then provide insights.
These insights can then be used to prepare documents for the next jurisdictions being entered at a fraction of the time it would take to search through the data manually. On the consumer side, an online sportsbook could use a Private GPT to allow bettors to ask questions on past team performance with answers generated based on the data held by the operator. Bettors can use the answers and insights to guide their betting decisions moving forward.
Private GPTs are a thing of the now
Fujitsu has recently rolled out a Private GPT with several customers across a range of industries, including online gambling. It sits on top of the company’s data warehouse with data being properly segmented. Permission can be given to specific teams and even individuals so that different users can access different data sets.
Those using it are already seeing the value they can extract from their data skyrocket, and without any major migrations or integrations required. The way we see it, our Private GPT really is the key to unlocking data which ultimately is what allows operators and suppliers to achieve long-term sustainable success.
