On December 5th 2023 we hosted our first Gen AI Summit, bringing together people across fintech to discuss the development of Gen AI within the sector. If you weren’t able to make it, or just want a recap, you can read all about it below.
The breakfast
We began the event with a breakfast, but we’re not going to write paragraphs on the quality of the croissants (though they were good). During the meet and greet guests had the chance to view demos from three Generative AI innovators.
Quick Action: A no-code platform to harness the power of Generative AI. Quick Action provides a secure environment for business teams to experiment and embed AI into their workflows.
Kloo: Next-generation accounts payable. Empowering businesses to unlock AI-powered automation that seamlessly integrates with ERP, for maximum control and efficiency.
CommsCoach: QA and compliance excellence. Built to help businesses have better quality customer conversations, so they can protect their bottom line by minimising risk and improving customer experiences.
Gen AI at Modulr
Modulr’s very own Gen AI lead, Scott McGregor, kicked things off, explaining the Gen AI tools we’ve created to streamline time-intensive tasks while keeping data secure. The GenBot email mailbox, and our Slack application ModBot , act in a similar way to ChatGPT (using OpenAI’s GPT4 model), but are only accessible to Modulites and keep all information fully private . They can be used to draft content, query documents, summarise text and more.
We also have a new AI tool for the partner onboarding process. It assists analysts review documentation submitted by potential partners and provides evidence to help determine the maturity of their frameworks and processes. Of course, this still requires supervision from the team, but a task that usually takes one team member a full day is reviewed by the ‘Onboarding co-pilot’ in under two minutes.
Sean Robertson, our Director of Data, then demonstrated a new velocity controls prototype. This solution analyses past transaction volumes and uses machine learning to forecast future activity, helping set suitable velocity controls per partner. Using generative AI, the tool also provides insightful commentary and guidance for our oversight analysts, enhancing decision-making and assisting Modulr’s risk management approach.
The prototype was conceptualised during our recent hackathon. This brought together 30 members of staff across tech, ops, legal, sales, product and more. It was far from the only idea however, and we’re planning to build out some of the other suggestions that came from that workshop, and we’ll hold another hackathon in the future.
Since the introduction of Gen AI, 81% of Modbot users reported that they’re more productive, while 84% think they complete tasks quicker. First signs have been very positive, but we’re confident that our best uses of Gen AI are still ahead of us.
Panel discussion
Fintech commentator David Birch was on hand to moderate. He began by alerting the audience to the existence of DaveGPT, his own chatbot for…well reasons yet to be fully discovered. But it acts as a good demonstration of how interest has grown over the past year.
David was joined on the panel by Modulr CEO Myles Stephenson, FIS Receivable Sales Manager EMEA Benjamin O’Connor, EY’s Payment Strategy and Technology Lead Anu Widyalankara and AWS Venture Capital/Startup/PE Advisor Leon Rees.
Here are some of the topics that were discussed:
Streamlining payments
How can Gen AI best be applied to payments? Benjamin O’Connor from FIS gave the example of empowering teams to simplify operations, without needing technical experience: “We’ve been using predictive AI for a long time in modelling, but with the introduction of Gen AI, our staff don’t need a computer science degree to interact with it.”
Leon Rees from AWS talked about the huge leaps that can be made in the customer journey: “Beforehand payments were about manually moving data files. Then we decided to digitise that, which was ugly and clunky, before progressing by making it all about the customer journey. Introducing Gen AI, we can outline what the right payment experience looks like and train the tech to take the customer through that journey to get the outcome they want.”
Security
David Birch raised some reservations over security and good practice, considering ChatGPT’s infamous ability for inventing information and the potential for adversarial AI. For Modulr CEO Myles Stephenson, this was why supervision was still key to the use of Gen AI within payments.
For EY’s Anu Widyalankara, fraud was something that could be combatted by collaboration from regulators, payment providers and other parties, such as social media companies. Through collaboration, Gen AI could improve security and fraud monitoring, not weaken it.
Leon Rees made the point that the data useful for reducing fraud has always been available to banks, but the problem is it’s never been ‘labelled’ in that way. That’s what Gen AI can help with, signalling the data that can make fraud monitoring much more efficient.
Bots vs people
The discussion then moved to where the use of Gen AI should and shouldn’t be used, or more specifically, what can humans do that Gen AI can’t. Anu Widyalankara made an interesting comparison.
“I think we’re moving towards not just humans talking to bots, but bots talking to other bots. When you think about when vegan meat substitutes became popular, the main positive feedback people gave was that if they couldn’t tell the difference, they didn’t care whether it was meat or not. I think we’ll have a similar response to Gen AI. We don’t care what’s behind the scenes as long as we get our desired outcome.”
Leon Rees continued that “if you have this high level intelligence capable of problem-solving, it makes more sense to point two of those things at each other and get them to talk, instead of a human talking to it.”
Audience questions
The above only scratch the surface of what was discussed, and we didn’t have time for all the closing questions from the audience, but we’re delighted there was such engagement. Questions varied from ethics in Gen AI, combatting inherent bias and whether in the future customer support should be needed at all.
Questions like these are what’s needed to further Gen AI and its use within payments. And, it was great for us to host a forum where they could be raised and discussed. It’s something we hope to do again in future.
Be sure to follow our social channels for updates on any other Gen AI events, and reply to this email if you’d like to chat with our team.