‘We need to go head to head with US tech businesses’
The ex-army man behind start-up Adarga tells Matthew Field how AI may be key in the global fight against terror
‘How on Earth did I end up running an AI company?” Robert Bassett Cross asks as he sits down over coffee. For the former Army officer and special operations expert, it’s been an unlikely journey. He spent years fighting in Iraq and Afghanistan before launching what could be the UK’S next bet in big data and artificial intelligence.
Unsurprisingly, Bassett Cross does not come across as your typical tech chief executive. The Silicon Valley uniform of sandals and T-shirt won’t do for the ex-military man, who later worked as an investment banker at JP Morgan. He cuts a disciplined figure.
His new venture – Adarga
– uses AI to change the way intelligence agencies and defence companies tackle terrorism, helping crunch vast amounts of data to identify suspects, hotspots and dangers even before they occur.
Adarga is a London and Bristol-based start-up. The venture is just emerging from stealth-mode, largely keeping its technology – which is already being used by the UK Government – under wraps as it fine-tunes it. The aim is to use big data to head off terror threats before they even begin. The same technology can also be applied in financial analysis, the insurance industry and a host of other areas, he says.
Founded in 2016, it now has 26 employees and is understood to be in the early stages of a fundraising that could raise tens of millions of pounds. Bassett Cross has real-life experience with using data to hunt down terrorists. He was awarded the Military Cross for his service in the British Army, where many overseas operations relied heavily on intelligence and data gathering. US and UK forces were then using software from a company called Palantir to do their data analysis. But something was missing, he says. “We started applying technology early on,” he says. “But it struck me at the time this was deeply inefficient. We were missing things on the cutting room floor – valuable information.” Of course, in a war zone, missing key intelligence can cost lives. But the sheer volume of intelligence was overwhelming.
Bassett Cross joined the Scots Guards in 1999, later serving in covert operations against terrorists. “A few days after 9/11, I was on a plane to Tajikistan. I didn’t come back [from active duty] for five years.” He finally left the Armed Forces in 2007, joining investment bank JP Morgan where he worked on major deals such as the listing of Just Eat. One of his first tasks at JP Morgan was, in fact, to return to Afghanistan to work on a deal for a prospective new gold mine. He jokes that his wife said she thought he joined JP Morgan to get out of war zones: “I guess they thought I was the only one crazy enough to go.”
Two years ago, he founded Adarga, taking on his first two employees, both coders from major defence firms.
While at JP Morgan, he began encountering similar problems to those he had seen in the Army: too much data and too little time.
“I felt I didn’t solve the problems during my time in the Army,” he says. “I hadn’t finished the job. I had the same problems at JP Morgan. Why are we reading slides at 3am? We have too many emails, too many documents, and yet we don’t read enough.” His start-up aims to engage in the big data dark arts that have yielded financial success for US rivals.
The idea of using AI to sift through reams of data is not new. What Bassett Cross hopes is that the technology can be used to trawl vital information that might otherwise be missed from millions of news sources.
Clients, largely in government and defence, can also plug in their own data sources like documents, PDFS and emails. The technology for now is specialised to his own history, dealing with issues of national security.
Bassett Cross hopes Adarga’s pedigree will set this technology apart, and possibly give the UK a future challenger to the US data behemoths. The company has received support from Benevolent AI, one of the darlings of the UK tech scene. Benevolent AI develops artificial intelligence software that analyses molecules to find new cures for diseases by looking at all the data available for use in pharmaceutical drug discovery.
It is a major shareholder, and has provided not financial backing but support in kind and technical help.
Having seen its software, Bassett Cross knew what he needed. “I met with Ken Mulvany [Benevolent AI’S founder]. He was showing us all these molecules, but I was seeing terrorists, I was seeing organised crime, I was seeing Russian interference.”
With Adarga’s technology, users can build up a “knowledge graph” about subjects and targets. Take a terror suspect: using data already available it can surface information about who they are connected to, where they live, who they have been in contact with and where they work, building up a web of details.
“We had all the data points for the last terror attacks on London Bridge,” Bassett Cross says. “We knew his guy, we had his credit card, we knew the hire company, but nobody was able to stitch it together quick enough. We think this technology could string these points together in a nanosecond.”
Adarga is trialling its products with defence companies, governments and security services. The firm is listed as a supplier on government websites.
Of course, the use of artificial intelligence for defence forces has
‘I felt I didn’t solve the problems during my time in the Army. I hadn’t finished the job’
been thrown into sharp relief in recent months. In the US, tech companies like Google have seen staff revolts over the potential use of AI to train drones.
Google was providing AI assistance as part of Project Maven, a drone programme for the Pentagon. Thousands of employees wrote in to oppose the deal, forcing Google to write a new charter to block work on weaponised AI.
Bassett Cross suggests this kind of thinking is misguided. “That fear is unwarranted,” he says. In the military, the kind of checks and balances needed before any action mean there is a human in the decision-making process at every stage. Even if there was further automation, there would need to be accountability, he says.
There are also fears about bias in artificial intelligence. For example, computer vision software used to recognise images has been known to confuse people of colour and women more than white people and men.
But humans exhibit even worse bias than a machine, he says, adding: “It’s a misguided fear.”
Perhaps this kind of ruthless thinking is what is needed to give the UK its first artificial intelligencedriven defence start-up.
“Nobody in the UK builds for global aspirations,” he adds. “We have too many potting shed businesses. We need UK and global talent. We need to go head to head with US tech businesses.”
Certainly, to take on the likes of big data giants like Palantir at their own game, the UK start-up scene will need plenty of funding, talent, and more than a fair bit of cut-and-thrust ambition.