In brief
Banks can only get value from AI if it runs on live, real-time data rather than yesterday's batch reports. Guillaume Aymé, CEO of Lenses.io, argues that customers now expect instant answers, and that fraud, cyber threats and AI-assisted service all depend on acting within seconds.
The tension is that banks carry 50 to 60 years of legacy systems and few engineers who know streaming data, yet cannot wait for full core replacement. Aymé's answer is tooling that sits on top of existing systems, plus trust, governance and multi-vendor flexibility, so AI is introduced safely, with humans still in the loop.
Questions
What does 'real-time data' mean for a bank?
Aymé describes it as the ability to act on data at the moment it is generated, which he calls a competitive advantage. ATM withdrawals, payments and fraud checks already work this way, and customers now expect the same from apps and AI.
Why can't banks just wait to replace their old core systems before adopting AI?
Jim and Aymé note that full core replacement is huge and slow. Aymé says Lenses sits on top of existing systems, finds the real-time data already there and lets teams build on it without moving or changing it.
What is Lenses and what does it do for banks?
Aymé, its CEO, describes it as a 'data operating fabric' that helps teams discover, check, govern and process streaming data. He says it makes building real-time applications easier so banks need not retrain everyone or hire an army of top engineers.
Why is real-time data so important for fraud and cybersecurity?
Aymé says teams may have only seconds to respond to a cyber threat, and slow fraud response can mean millions in lost funds and money laundering. He argues attacks are too fast for humans alone, so security agents need live data.
Should banks put AI agents directly in front of customers?
Aymé says the industry is recalibrating, because agents are hard to master in regulated sectors. The safer pattern he sees is AI assisting a human-run process, and Jim adds that customers could confirm any recommendation before it acts.
Facts
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Forrester predicted that by 2025 half of all businesses would use AI-powered self-service as their primary customer touchpoint, according to Jim Marous.
“Forrester predicts that by 2025, half of all businesses will use AI-powered self-service as their primary customer touchpoint.”
0:00:11 · Jim Marous · AI self-service · 2025
Lenses.io describes itself as a leading player in what it calls a data operating fabric for real-time data.
“We are one of the leading players in what we call a data operating fabric.”
0:01:33 · Guillaume Aymé · Company description
Guillaume Aymé says that after the financial crash, regulators demanded real-time insights into banks' positions, prompting modernization of systems.
“you could say that after the financial crash there was kind of increased pressure on cyber resilience regulation regulators were demanding for real-time insights on the position of banks”
0:03:28 · Guillaume Aymé · Financial regulation · After the financial crisis
Guillaume Aymé says open banking drove real-time data by forcing legacy banks to open their platforms and share data with third parties.
“open banking uh really had a great uh was a great driver of real-time as well having forcing on the legacy players to open up their platforms share data to third parties”
0:03:28 · Guillaume Aymé · Open banking
Guillaume Aymé says banks live with 50 to 60 years of legacy infrastructure.
“The banks live with this legacy of 50, 60 years of archaic infrastructure.”
0:08:25 · Guillaume Aymé · Legacy systems
Guillaume Aymé estimates that about 5 to 10 percent of bank infrastructure has been modernized to real time.
“Some of it, probably maybe about 5 % or 10 % has been modernized through real-time.”
0:08:25 · Guillaume Aymé · Bank modernization
Guillaume Aymé says a large US bank told him it had 400 projects in its modernization backlog even without AI.
“I was speaking with a large financial services, US-based bank. And they said even without AI, they had 400 different projects in the backlog”
0:15:46 · Guillaume Aymé · Project backlog
Guillaume Aymé says customers of Lenses.io generally get value within days of effort.
“generally we're talking about days of effort to get some value”
0:15:46 · Guillaume Aymé · Implementation time
Guillaume Aymé says Lenses.io sits on top of a bank's existing systems and discovers the real-time data already there without changing anything.
“it sits on top of an organization or a bank's existing investment. It's not asking to change anything.”
0:11:35 · Guillaume Aymé · Lenses architecture
Guillaume Aymé says slow fraud response can cost millions in lost funds and can lead to money laundering.
“fraud could be millions of lost funds if you're not responding quickly to fraud. Also can lead to money laundering and all sorts of other challenges.”
0:10:19 · Guillaume Aymé · Fraud
Guillaume Aymé says Lenses.io has more than 10 years in the real-time data space and has worked with large Fortune 500 companies.
“We've got more than 10 years in this real-time space, having worked with some of the largest Fortune 500 companies.”
0:36:44 · Guillaume Aymé · Company experience
Guillaume Aymé says the majority of Lenses.io customers are in financial services.
“the majority of our customers are within the financial services industry.”
0:36:44 · Guillaume Aymé · Customer base
Guillaume Aymé says a very large US bank shares real-time data with ServiceNow for cybersecurity ticketing.
“A very large US customer, banking customer, shares data with a big cloud provider ServiceNow.”
0:24:22 · Guillaume Aymé · Data sharing
Guillaume Aymé argues banks gain by sharing fraud and attack information in real time, even with competitors.
“everyone has to gain by sharing fraud and attack vectors even to their competition”
0:24:56 · Guillaume Aymé · Fraud sharing
Guillaume Aymé estimates that about 10 percent of the world's data is publicly available and used to train large language models, while about 90 percent sits behind firewalls.
“I think something like 10 % of the world's data is publicly available what these LLMs are trained on, Wikipedia, literature, and so on.”
0:26:43 · Guillaume Aymé · LLM training data
Lenses.io released a product called a replicator that moves and shares data in real time inside and outside a business.
“We've just released what we call a replicator that allows you to move and share your data, both internal to your business and external to your business in real time.”
0:27:23 · Guillaume Aymé · Lenses replicator
Guillaume Aymé says the safest pattern he sees is AI in the loop of a human process, rather than a human in the loop of an AI process.
“where there's AI in the loop, it's a human process. Rather than human in the loop, it's AI in the loop.”
0:29:45 · Guillaume Aymé · AI in the loop
Guillaume Aymé describes a Lenses.io customer using real-time voice sentiment analysis to tell call-center operators whether a customer is frustrated.
“there's AI systems that are analyzing the voice of the customer in real time. And that voice analysis is doing sentiment analysis.”
0:30:19 · Guillaume Aymé · Call center AI
Guillaume Aymé says Lenses.io is embedding AI into its own product to help engineering teams work faster.
“we're introducing ai into the product itself that's embedded in the workflow of some of these teams”
0:34:04 · Guillaume Aymé · Lenses AI
The Banking Transformed podcast is described by its host as the winner of three international awards.
“the winner of three international awards for podcast”
0:40:17 · Jim Marous · Podcast awards