Mimiro — the financial industry’s knight in Machine Learning armour

Him: “Would you like a coffee?”
Me: “Sure — thank you.”
<some time passes>
Him: “Actually, I didn’t know how to make a coffee here, so I just went out and bought you one.” <Thrusts Costa coffee cup in my hand.>

The exchange above is, almost verbatim, my first interaction with Charlie Delingpole, CEO and Founder of ComplyAdvantage, back in January 2016. For me, it captures precisely why Charlie is a great entrepreneur — he saw a problem first, resolved to solve it and found the best solution to the problem through ingenuity. What he did not do was brew a perfect latte and then walk the streets looking for someone who might need one.

Earlier today, the company he founded announced that it had received $30M of financing to continue its mission of building a machine-learning platform for analysing risk of financial crime. The round was led by Jan Hammer at Index Ventures who I have now known for a long time and has worked with some of Index’s best known financial technology companies including Adyen and Robinhood. I know I speak for the rest of the board and team at ComplyAdvantage when I say that I’m really excited to get a chance to work with Jan and see his huge experience come into play as Charlie and his team continue to build the company.

My partner Tim and I have been working with ComplyAdvantage since Balderton led the Series A investment back in 2015. Balderton is relatively unusual in being built on an equal partnership (link: https://medium.com/@suranga/why-equal-partnership-vcs-are-better-for-entrepreneurs-60910aa8b3eb) where every partner is equally incentivized (and responsible!) for every company we work with. In practice this can mean that more than one of us will be involved in everything we do from finding and investing in a company to sitting on its board and working with its team after investment. ComplyAdvantage was just this sort of case — Tim’s incredible background in the financial services industry has helped the company from a customer and commercial strategy perspective, while I have worked more on product and technology strategy, applying some of the lessons learned at Autonomy and blinkx on the challenge of building huge AI systems that work without bias and at the speed and replicability modern industry demands.

Not the same coffee cup as the one Charlie handed to me in 2016. But very similar.

Although very much a company that is visible only to a small population of buyers and users I think ComplyAdvantage is one of the most exciting companies currently being built in Europe’s burgeoning ‘fintech’ industry. Technology and open border financial policy have brought consumers and businesses an incredible plethora of financial products. It is easier today than ever before to find the right loan, move money around the globe, build complex international supply chains, sell to anyone anywhere and do it all with the ease of a click on a laptop or a tap on a phone. This simplicity of access has also been a boon to the bad actors. The world is full of those who seek to exploit this state of affairs to launder money, fund terrorist organizations, avoid tax and more. The financial services industry, taking this threat head on, has appropriately scaled compliance and risk management. Today, global finance organizations spend literally billions of dollars every quarter checking and re-checking the background of every entity and individual, the provenance of every deal or transaction and the cascade of impacts that each movement of an asset causes in this web of inter-connected activity. Much of this is manual. There are ‘farms’ of analysts sitting in warehouses compiling and comparing thousands of reports in the media in the hope of spotting PEPs (Politically Exposed Persons); there are legions of compliance officers in high-rise buildings starting at transaction logs, hoping to find patterns of unusual behaviour.

At its heart, Mimiro’s backend ingests millions of transactions, media reports and financial signals to build a highly accurate, up to date risk model of the global financial system and its actors

The ingestion of unstructured data, the analysis of millions of transactions in the hunt for the outlier patterns, the building of profiles and networks of influence and connectivity. These are the sorts of problems that are perfect for machine learning and that’s exactly what Charlie realised when he started the company. A serial entrepreneur who founded and ran The Student Room out of his teenage bedroom and then went on to co-found MarketInvoice after university, he had stumbled into the finance industry from the outside. Realising that the challenge of compliance was going to only grow and that existing solutions to the problem would never scale he had an inkling that machine learning approaches might hold the solution. An autodidact of phenomenal ability, he learned about the field rapidly and, realising that this was his way of getting a much-needed cup of coffee into the hands of his customer, started the company shortly after.

That was then and this is now. Today ComplyAdvantage is 160 people, serving 350 clients in 45 countries. This year, when the company’s business more than tripled in size, was driven partly by a huge growth into the US where they signed 100 customers in a single year.

Charlie Delingpole, Founder and CEO of ComplyAdvantage

Rapid adoption in a market notorious for long sales-cycles and complicated approvals has been driven by the fact that they just solve the problem better and faster. Banks, insurance companies and really any corporate that needs to decide whether to transact with a previously unknown counter-party can do so in seconds, via an API request, understanding risk on an instant, programmatic basis. Existing technologies are built on simplistic keyword matching where, especially in the case of common terms like names, means there are thousands of false positives that have to be painfully double-checked by hand. Mimiro’s intelligent approach that folds context into matching means they cut these by around 70% for most clients. And by having automated spiders and bots trawling the web for the underlying data 24x7x365, rather than human analysts, the system is up to date — sanction updates and changes in an individual’s status are reflected in the system in seconds and minutes, not days and weeks.

This technology and the products it powers may be invisible, but their impact will be felt by all of us. Next time you fly abroad, seamlessly withdraw money in a new currency, buy something that was built from components sourced from six companies, flawlessly share a bill with a friend or set up a new account to handle a new project — behind the curtains, making it happen fast, safely and with a full appreciation of the risks involved will be ComplyAdvantage and its engine.