What are the key AI trends you anticipate will play out in the financial services industry next year? The obvious one would be the progress of frontier models. The interesting challenge will be the ability of all of us to harness that power. We have already introduced a number of new products and we will scale those up quite nicely. Speed in the development process will be highly relevant here, as all of our engineers are using AI tools today. At the same time, we are developing a more comprehensive platform for them to test codes, promote them, monitor and optimise them. That will increase productivity and help us advance our product roadmap. In fact, we have already improved our productivity by 25 per cent in the last year, and the target is another 25 per cent in the coming 12 months. Individual productivity due to AI has yet to translate into organisational productivity at most enterprises. Isn’t that a concern? We said that we will have $25 million of savings from AI. While that is a small number relative to our total cost base, it is a very specific number. A lot of that was from technology and particularly around testing, where we have been able to automate so much of what was previously manual. We have also told investors that we will be reinvesting all of that because of so much change happening. Broadridge is part of Project Glasswing, Anthropic’s collaborative cybersecurity initiative to test its advanced AI models such as Mythos. How has that helped you strengthen your cybersecurity systems? That is an area where we are investing a lot. As you know, we have to use AI to fight AI. One of the big things that Mythos can do is take vulnerabilities that were not deemed critical and string them together to make something that is critical. We, like many other companies, are using AI to go back and clean up the backlog of non-critical vulnerabilities. We have already reduced our non-critical vulnerabilities by 80-90 per cent in the last few months. We are also using AI to radically improve the speed of patching — the process of finding a vulnerability and fixing it within a certain amount of time. That is a safe amount of time before it would likely be exploited. That time is shrinking dramatically, so we are using AI to detect and remediate faster. The three pillars are speed, cost and security. What is the one thing about Mythos that should make executives nervous and excited? I think the thing that is making everyone nervous is this ability to take multiple things that are not critical and put them together to create something that is critical. That has created a lot of work for all the leading institutions to go back and resolve those. However, our ability to use AI to remediate things more quickly, identify and patch things faster, and dramatically accelerate our whole security process is something that I am really proud of. Does that change the type of talent you would like to hire? That is interesting because it does. It is actually fewer people doing the work and more people designing the systems that will do the work automatically. There is a lot of intelligence and a lot of domain knowledge needed now. The skill set is shifting from pure-play delivery to engineering design and solutions. That is where we are seeing the transition. While using AI internally, how do you minimise the use of tokens? All the companies that I know of are looking at how they engineer things — using the simplest possible query and having software that automatically uses the cheapest possible model; perhaps sourcing models from China and using open source. There is a lot of engineering going into minimising the number of tokens being used. When I look at our token usage, it is pretty immaterial relative to the savings that we are getting, and so the return on investment is very good. What has been the role of your Indian centre in your AI journey? We have a part of our global leadership here. We have a platform that pulls together all of our applications with a common data model and a common set of APIs, which is great for applying AI. The leadership for that is coming from our team here.
——————————————————-
Click Here For The Original Source.
