From how a tiny faction of AI zealots may be pushing us too fast, to the importance of the EU AI Act, to philosophy, we had a wide-ranging chat with Jared Browne, Fexco Group head of privacy & AI governance.
Since its founding by Brian McCarthy in the perhaps unlikely location of Killorglin, County Kerry, in 1981, Fexco has grown into an international operation doing business in over 50 markets and employing some 2,600 people. Like all organisations, Artificial Intelligence (AI) is impacting its internal operations and external business, so it was a timely opportunity to speak to its head of privacy and AI governance, Jared Browne.
A philosophy graduate, and qualified lawyer, Browne is also a member of the EU General Purpose AI (GPAI) Code of Practice working group, so also good timing as the EU’s general practice AI code comes into force this weekend.
Fexco started out as a financial services company, specialising in foreign exchange (FX), and that is still at its core, but today it has evolved into many other areas, as Browne explains. “It’s a complex conglomerate at this stage, across about 15 companies. There’s an awful lot of extra things going on now, and I’m always surprised really at the willingness to try out new things.”
An “opportunity-led organisation”, according to Browne, these days Fexco broadly breaks down into three sector – its payments and FX division, a managed services and advisory division, and a separate services division that includes things like property management businesses.
And it is very much a global operation. Still headquartered in County Kerry, in Ireland, it has a footprint in the UK and Australia, as well as New Zealand and the pacific islands, and further afield. Just days before we spoke, Fexco had almost doubled its retail FX business in the UK, with the acquisition of Sainsbury’s Travel Money.
“As well as its legacy business, today there is very busy product and innovation team which is trying to basically figure out what the future is going to be, moving away from what we’ve done in the past and into where the shifts and the changes are going,” says Browne.
AI inside and out
As with most organisations, he says Artificial Intelligence (AI) is getting much attention, in terms of how Fexco brings its own staff forward, and also in terms of how to respond to market need in the AI space.
Internally Fexco has a programme called the AI Launchpad where it is proactively doing workshops sessions with all business units, to ensure they have a level of comfort using AI tools in their roles. “We’ve also done AI literacy training from February onwards for all of our staff, and for two years now we’ve had a Central AI council, to try and figure out the direction of AI in Fexco. That meets every month to assess product ideas, and to look at all the various AI initiatives that are going on.”
Externally, Fexco is working on developing appropriate AI product offering for clients. It has been offering its Smart Assist service for three years now, a generative AI platform, “effectively using Azure Open AI”, says Browne. “We have sold this to a number of government clients already, including SEAI and the Residential Tenancies Board, and what this is doing is trying to address the boring problems. The boring problems tend to be the ones where there is actually that thing called ROI (return on investment) attached onto the end, which you’re not always going to get in an AI use-case.
“There are plenty of AI solutions running around in search of a problem, right?” says Browne. “But Smart Assist is addressing issues like taking a client’s knowledge database, using that, and then when the thousands of emails come in from the public, Smart Assist will generate responses to all those emails by basically plugging into the knowledge database.”
“We’ve built this meticulously and slowly, with the security and accuracy bit built in, simply because generative AI in its native state, you wouldn’t unleash it on the world. If you use things like retrieval, augmented generation and contextualised responses, and meta prompt management, you get this kind of locked down environment where you get the power of generative AI, but only within this walled garden of the client’s knowledge database.”
With Smart assist Browne says clients are getting their e-mail response time down from 28 minutes to three, including full human review – a “boring” problem that certainly needed fixing considering the onslaught of emails all organisations face today.
The human review element is one close to Browne’s heart. “Our approach is human first rather than AI first, so there’s never been a consideration of trying to remove roles. We’re looking at how we can augment and improve the efficiency of all the staff that are there. In our managed Services division where we do a lot of customer service work for government agencies and large companies, all of the staff now use Smart Assist next to them every day in their work. And this is prompting them with answers and information live. This is something that we put a lot of work into – getting that interaction right because the way an AI and human interact needs some fine tuning.”
Human-first, AI-assisted lending
Elsewhere, Fexco’s 50-50 joint venture with 17 of the larger credit unions in Ireland, Metamo, recently launched the first AI lending product in Ireland, says Browne, which helps credit unions speed up the decision-making process for loans.
“The loan officer really is still the main player in there, so it’s just one of the pillars, using AI to assess repayability, finance, interest rates, all of that. And so that is something that is speeding up the lending process for credit unions,” says Browne, who believe the credit union space is one of great opportunity.
“With the credit unions, there is the small matter of still being able to meet a human being, or talk to one on the phone, and there’s an awful lot of money locked up as well in the credit union sector that that isn’t being used. That’s a big opportunity space, you’re going to see more products and services coming, in conjunction with Metamo.
Browne’s philosophy of AI
Browne works in AI, and he believes that agentic AI is going to transform much of what we do in the coming few years. He’s not anti-AI per se, but is concerned about the rush to implement, driven by the few. In his own words “once a philosophy student, always a philosophy student”, and today he finds his very particular set of skills serve him well is an AI-obsessed world where we’re all grappling with the ethical quandaries.
“What someone said to me once about philosophy – and it always stayed with me – is that it leaves you chronically unimpressed at the world, but in the right kind of way,” he laughs. “In that you get those critical tools to analyse everything and you’re less willing to take things on word, right?”
Far from simply the “human in the loop” approach, Browne is a strong proponent of human-first.
“Why would we even bother with technology if it wasn’t to serve human beings interests?” he asks. “What I find endlessly baffling at the moment is that we seem to be developing a technology now that is going to be telling us what to do, and which seems to have to inevitably develop, no matter what the consequences for humans. And of course, that technology is called AI.”
“I just do not get it,” he continues. “From the beginning of time, we’ve built tools because we needed to achieve something. And what’s different this time? I think it is absolutely essential that we can build this strictly in terms of what humans need to improve our lives, and that the most vulnerable people in society would not be affected by it, indeed that it would also actually serve the most vulnerable people.”
Above all, Browne worries about the sheer rush to develop. “I think it is all moving too fast and there’s a timeline from four or five companies in San Francisco, which is that they need a return on their investment next year, for their investors. Whereas I think the human race needs about 10/15/20 years to make the sorts of changes that they’re telling us are coming very, very soon – the changes to employment to the way they work.”
“And it’s not just the work environment. States and governments need to understand how is all of this going to affect income tax revenues, will there be less people working? Will you have unemployment? Will you have unrest?”, he says. “So I’d have a lot of concerns about that direction, and this very tiny faction that’s pushing that agenda. I feel that governments really do need to insist on the social good side of this technology, and not be led too much by the market.”
EU gets its Act together
It’s why he’s impressed with the European attitude to the impact of AI and its efforts very early on to tackle them head on with the EU AI Act.
“As a proud European, I’m a big fan of the AI Act,” says Browne. “I think what Europe is getting right is that we need to slow down AI in order to speed it up again, to insert a governance layer into the middle of the whole thing that makes sure that there’s going to be some safety and risk awareness in here, so that we actually get the benefits and minimise the harms.”
“The EU doesn’t get everything right, but I I think they did get this right in that they want to win the long game, not the short game. If people in the EU get to trust AI because they can see it can safely run huge public services, then you’ll get real adoption, more purchasing, more buy-in and more economic success in the long term.”
“If this is as big as everyone says it is and everyone says it’s big, the Americans, the Europeans, the Chinese, why would you be rushing it? Why would you not be meticulously trying to get it right and make sure that it succeeds? So the AI Act – and I’m sure it’s not perfect – is at least trying to do that careful, slow, precise, long road to success for something which is going to be vital. We wouldn’t accept bad air in our cities or dirty water. When it comes to aviation safety no one argues about planes being safe before they get off the ground. These are totally accepted things. That’s exactly where AI is going to have to land.” Hear, hear.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
