PureGroup's SalesAI has been shortlisted for Investment Week's Marketing and Innovation Awards under Best Use of Data and Open Innovation Award categories
Is a global battle about to break out between AI-powered digital sales technologies and human sales personnel? Or are they all part of the same fast-evolving team?
Patrick Murphy, founder of UK-based tech company PureGroup, is betting that humans need empowering, not replacing, by technology and that this is especially true in asset management. His newest product, SalesAI, uses natural language generation to help sales teams deliver insights to clients about fund performance and portfolio construction versus the fund's nearest competitors, in the light of market trends.
"For us, technology is about enhancing the human sales resource by supplying timely, personalised insights based on vastly more information than humans can curate themselves," he says.
But today's deluge of financial content means ‘less is more'. "There's an understandable desire to give increasingly sophisticated clients more detail, more reports, more information. Asset managers should do the opposite and use technologies to aggregate, summarise and provide as much insight as possible in the fewest words," Murphy says.
War on words
His London and Lisbon based team of young engineers developed SalesAI in a six-month sprint in earlier 2019 and it is now in use by clients. "Essentially it overlays AI on multiple fund and market data sources - currently over 12 trillion data points within our data base and rising fast," he says.
The number crunching has to be on an impressive scale to sieve out the most compelling, personalised stories: "So if there is growing demand for UK equities, how did our fund perform in that rally period versus the competitors the customer thinks most important?" Personalising insights also means taking account of dimensions such as how long the client has been invested in the fund.
"But the real trick is to boil down a potentially huge range of data-driven insights to the 3-5 most compelling statements that sales can provide to each client," says Murphy. Sales can then engage clients without waiting for input from analysts.
Murphy says this is where SalesAI's most unique aspect - client-configurable natural language generation - comes in. Natural language generation, a key plank in the artificial intelligence revolution, is essential because each sales statement must ‘look and feel' as if it has been written by a human analyst.
But language isn't just human, it's culturally distinct. Murphy says some sales teams are succinct, others more expressive; some may wish to focus on performance while others wish to highlight more complex sales messages; and all must keep compliance teams comfortable.
SalesAI's natural language generation is therefore configurable to each client's tone, though it can also be deployed instantly in a ‘best of breed' style. "The text we write should reflect each asset manager's brand values," he says, "and it's wonderful that this is now technically possible."
Murphy says the approach will help asset managers redeploy analysts. "They no longer want expensive talent pulling data and running Excel spreadsheets on a monthly basis for sales and client support teams when it could be adding value elsewhere," he says.
It's anyway near impossible for human analysts to supply timely, personalised intelligence. Murphy cites a global asset manager with, say, 60 funds promoted by 30 sales people. Each sales person might have 250 contacts, each with their own investment needs and perception of the competitor landscape. Working out the most compelling story for all these contacts might take nearly half a million unique combinations of data - countless years of work if each combination took as little as an hour to compute.
By contrast, Murphy says SalesAI's computations are so fast that in early demonstrations, clients thought they were reading their own analysts' statements copied and pasted from their website - and weren't impressed. "So we slowed the app down by one second and gave it a visual loading-wheel icon so users realised the statements were being created in real time entirely independently and from current data." Humans need to see machines sweat.
Speed generates more than efficiency. "Putting a real-time analyst in the back pocket of each sales team member so they can offer instant, personalised insights rather than slow, generic information will help our clients grow assets," says Murphy.
His product sounds like it was designed for sales crews in lock down. "Wherever you are - any time, any place - you can get asked about a product and its competitors and you will be able to respond," Murphy says.
Sales people have always worked semi-remotely. But Murphy thinks the crisis has concentrated management attention on making remote working more efficient by breaking down internal knowledge silos and "liberating insights from all subject matter experts and their data."
SalesAI's data agnostic architecture means that in the future it could embrace deeper, richer sets of information about client preferences and behaviour. He says asset managers have "a view of data that is small in scale compared to the big data perspectives in other industries and are only scratching the surface of what is possible."
But he doesn't think AI-generated insights will sell products on their own. "Replace the sales person? A good sales person can recognise a sales hook a million miles away and technology can now help them find that hook quickly. But it can't replace human intuition about the client's position or human empathy," he says, pointing out that big tech - Google, Facebook - still has armies of sales people.
Investment Marketing and Innovation Awards 2020
This year's ceremony takes place online on Thursday 16 July. Click here to find out more.