For decades, our database-connected softwares have had a “download as Excel sheet” button. In the last year, industry analyst Vinod Jain, of Datos Insights, has noted the rapid spread of on-page chatboxes, staffed by articificially intelligent bots, ready to serve you with specific data that you ask it for:

“It could soon be that in 95 percent of the cases, you wouldn’t be able to sell an application unless it has an AI feature in it.”

It could be time to start chatting with the data on your computer screen, much as you would chat with a human colleague. 


Factbox: AI capabilities fit well for post trade
• With artificial intelligence being used for so many different applications around us, we’d be excused if its lines look blurry to us. In post-trade processing, one utility is its ability to extract data from semi-structured formats such as confirmations in print or pdf, into our fully structured data bases. This strength, in learning to recognise patterns, appears to be one of the driving forces enabling private-asset servicing in increasingly efficient-enough ways. Filing transactions into accounting programmes, based on just a scan of a receipt, could be a similar example from close to our daily lives. 
• Other useful functions include the speech-to-text capability that many people use to set the cooking timer on their phones, and – more recently – the growing use of context-aware chatbots, spearheaded most famously by ChatGPT. (Now, combine that speech recognition and context awareness together, and you can have a voice chat with a bot that gradually gets what you are asking for as you speak, and inches towards the precise answer in collaboration with you. What better colleague? Or call it a tool.) 

Now, this is where Vinod Jain’s prediction comes in. Exploding amounts of data flow through our various systems, and a growing number of financial institutions are looking into how they could extract value from their existing data, possibly to produce externally marketable data services. And on the subsequent question of how the user should interact with that data, the answer could be ‘just chat with it’.  So, while it has been normal for decades to extract data as spreadsheets, to slice and dice further on your desktop, the future may well allow us to ask verbally for precisely what we seek. 

“It is a must-do that there is an AI-based dialogue box on the user interface, ready for interaction. It cannot be hidden away somewhere.”

So, is there any big don’t? Vinod Jain sounds certain that it would be unwise to charge extra for such AI functionality, at least at this early stage. 

“Clients will want to see it but not pay for it – and internally, would you run into a risk of having to set up transfer pricing between departments?”

A big transition

Vinod Jain speaks with PostTrade 360° in connection with the recent release of Datos Insights’ yearly capital market trends report, noted by us here. The report’s post-trade observations included manual processes pushed to their limits, tension over the American shift to T+1 settlement, as well as massive shifts in technology and operational risk management. Still, on our question where trend insider Vinod Jain has, personally, perceived the most specific surprise element this past year, his thought zooms in on that little chat icon – together, of course, with the big transition it represents. 

One advantage with artificial intelligence is how it can learn to interpret different ways of asking for the same insights.

“This is very useful from an operations perspective. One person may ask whether a specific trade has failed, the other could ask for a lists of failed and successful trades processed at 3 pm today – though they are both looking for information about the same trade.” 

Many possibilities

Another benefit for data-intensive businesses, in Vinod Jain’s eyes, is a greater freedom in designing new services. Historically, it was necessary to design any data feed on the basis of what data was comfortable to retrieve. Increasingly, the artificial intelligence tools allow for data extraction that is driven more explicitly by the value of the outcome for the user – not hampered so much by the practical challenge of drilling down to it. 

Yet another use case, as our operations increasingly lean on programmed applications, is that a chatbot such as ChatGPT can shorten our path to functional solutions, when we start from our high-level idea of what our programme should do. 

“It can give you a jump start to build something you don’t have. Let’s say you want to learn about tokenisation; you can say ‘hey, AI, can you help me understand those big numbering sequences over there’.”