The Virtual Agent: NLP in Wealth Management

The Virtual Agent: NLP in Wealth Management

NLP has many use cases in consumer banking and is gaining adoption in wealth management. In my report, The Virtual Agent: Natural Language Processing in Wealth Management, I look at some of the more popular use cases for NLP, including for chatbots/virtual assistants and biometric identification, as well as the more cutting-edge applications, like for advisor matching and more complex virtual assistant processes.

IPSoft and Personetics are two vendors making huge strides in the field of NLP.  IPSoft is working with SEB and Personetics is working with BRD, a subsidiary of Groupe Société Générale, to explore innovative use cases for NLP.

Enterprises looking to explore NLP should consider whether a solution has a process ontology that builds best practices, works across multiple languages, detects formality, and perceives when a human should get involved.  Best practices should be tested on internal use cases before applying NLP technology to external-facing problems.

In the next 12 months, it is likely we will see many of the largest consumer banks that already have a virtual assistant rolling out new uses cases for their virtual assistants within both their consumer banking and wealth management arms. Other banks that do not yet have a virtual assistant or chatbot offering will be racing to catch up.

New Year New Tech New Research

New Year New Tech New Research
In your new year resolutions, did you pledge to understand more the technology that scares you? Or at least the one that some people (aka analysts like me) claim will replace you? If the answer is “No” and you are working in the field of Investment Research, whether producing, consuming or distributing it, then you may want to read our latest report Start Coding Investment Research: How to Implement MiFID II with Robots and AI.

I get paid to write research on fintech so theoretically I am not the tech scared type though I am the first one to control screen time at home. I know we have more and more competition from free research you can all find at your fingertips on the internet, and from cheaper research that leverages outsourced resources crunching a lot of data, but so far we are keeping up probably because our clients think we provide insight that those competitors do not provide yet.

I know however that we have competitors that have technological platforms that distribute their technology in a more user-friendly way with podcasts and fancy databases, that write their research in a more automated way and that you can consume easily because you pull the information with selective search technology that knows what you want and how much you can pay for it.

So before the holiday season, to make sure we were all going to start this new year with the right information in hand, I did look into what artificial intelligence and robotic process automation tools will be doing to research; not exactly my kind of markets fintech research, but more specifically to Investment Research, those written recommendations about equity or bonds or macroeconomic environments to help the buy side make investments.

The result is very honestly scary and exciting at the same time. These new  technologies are maturing at a time of big regulatory change in Europe, MiFID2 is finally kicking in and that means the unbundling of investment research cost from the execution costs the brokers and banks charge their buy side clients. Some buy side will keep using them and be happy to pay that fee, some clearly will start looking at other solutions that will have to propose a different business model provided by banks or by new market players, based on technology.

In our recent report we do look exactly at that: new business models and live case studies that have already been implemented in investment research production, distribution and consumption. Enjoy.

Introducing The Cognitive Advisor

Introducing The Cognitive Advisor

Last week I published a report on a topic that has interested me for some time: the application of artificial intelligence (AI) technology to the wealth management business. To date, neither Celent nor its industry peers have written much about this topic, despite clear benefits related to advisor learning and discovery. This lack of commentary, and the industry skepticism that underlines it, reflects successive waves of disappointment around AI, and more recently, competition for research bandwidth from other areas of digital disruption, such as robo advice.

Another inhibition relates to taking on an industry shibboleth. How to reconcile AI or machine intelligence to the hands on, high touch nature of traditional wealth management? This challenge is real but overstated, even when one reaches the $1 million asset level that has defined the high net worth investor. Indeed, the extent to which wealth management is a technology laggard (in general, but also when compared to other financial services verticals) highlights the opportunity for disruption.

In particular, AI offers a means to circumvent the dead weight of restrictions presented by antiquated trust platforms and other legacy tech, a weight which reinforces advisor dependence on spreadsheets and other negative behaviors. As is set out in the report, it is precisely the combination of new behaviors and technologies that can help surmount the finite capabilities of the human advisor.

Improving Operational Efficiency in KYC-AML Using AI solutions

Improving Operational Efficiency in KYC-AML Using AI solutions

Regulatory scrutiny and growing cost pressures are severely impacting Know Your Customer (KYC) and Anti-Money Laundering (AML) operations of financial institutions. Discussions with several banks have revealed they are finding it hard to keep track of constantly evolving regulations, interpret and implement global regulatory changes at a local operational level, collect and refresh information from numerous sources and systems across different businesses and jurisdictions, manage and analyze growing volumes of structured and unstructured data to identify patterns, networks, and beneficial owners, while containing costs amidst a difficult economic environment.

Banks so far have looked to address the challenges by hiring more staff, as traditional rule-based KYC-AML technology necessitates significant dependence on manual efforts. But too much reliance on manual efforts can make the process costly, error prone, and inefficient. Banks therefore need to think out of the box and consider new and innovative solutions to alleviate operational and cost pressures.

Adoption of Artificial Intelligence-enabled solutions could be one way to mitigate current challenges, increase efficiency, and reduce costs, as they can not only automate significant parts of operations but also offer superior insights through advanced capabilities for analyzing structured and unstructured data. In a new report, Celent discusses several challenges plaguing financial institutions’ Know Your Customer (KYC) and Anti-Money Laundering (AML) operations, and proposes how Artificial Intelligence (AI) enabled solutions can help in addressing them. This report was commissioned by NextAngles, an Mphasis Fintech venture, while Celent kept full editorial control. The report is available for download here.

Proof of artificial intelligence exponentiality

Proof of artificial intelligence exponentiality

I have been studying Artificial Intelligence (AI) for Capital Markets for ten months now and I am shocked everyday by the speed of evolution of this technology. When I started researching this last year I was looking for the Holy Grail trading tools and could not find them, hence I settled for other parts of the trade lifecycle where AI solutions already existed.

Yesterday, as I was preparing for a speech on AI at a conference, one of my colleagues in Tokyo forwarded me an Asian newswire mentioning that Nomura securities, after two years of research, would be launching an AI enabled HFT equity tool for its brokerage institutional clients in May –  here it is: the Holy Grail exists, and not only at Nomura. Other brokers have been shyly speaking about their customizable smart brokerage, e.g. how to use technology so that tier5 clients feel they are being served like a tier1. Some IBs are working on that, they just don’t publicly talk about it.

Talking to Eurekahedge last week I realized that they are tracking 15 funds that use AI in their strategy, I would argue there are even more than that because none of those were based in Japan (or Korea where apparently Fintech is exploding as we speak).

All this to reiterate that AI is an exponential technology, ten months ago there were no HFT trading solutions using AI, and we thought they were a few years away but no, here they are NOW. And the same with sentiment analysis, ten months ago they were just a marketing tool, now they are working on millions of documents every day at GSAM. Did I forget to mention smart TCA that’s coming to an EMS near you soon?

Stay tuned for more in my upcoming buy side AI tools report.

Being smart with artificial intelligence in capital markets

Being smart with artificial intelligence in capital markets

Artificial Intelligence (AI) is the new buzzword to talk about on the street. Financial institutions need to embrace AI, as we have explained in our January report, or else they risk to lose competitiveness or be coded by the regulators more than they can do it themselves.

I am in NYC next week to share Celent’s view on AI for capital markets. A little preview for your here.

Today we are at a crossroad where data scientists have the computing power, the alternative mind-sets to search and the willingness to look for narrow AI solutions, not the wide AI brain that we should get to in 2030 according to experts. This enables vendors to come up with amazing solutions from Research Scaling with Natural Language Generation to Market Surveillance/Insider Trading with Machine Learning Natural Language Processing or even Virtual Traders via Deep Learning of technical analysis graphics traders look at to take decisions.

The amount of data available is another big driver for the rebirth of AI, and regulators are looking at ways of accessing that data and using it. This is borderline what my colleagues would call RegTech, and it’s coming.

Our Q2 agenda reflects our understanding that you want to know more about AI: we will share ideas on solutions for the buy side, for exchanges and for the sell side. But in the meantime I hope to bring back some cool ideas from the big apple, hopefully also from the secretive quants working in the dark Silicon Alleys.

Most of the vendors I have profiled are specialists’ boutiques, but the cost of such research is however so enormous that generalists are trying to productize their fundamental research for various sectors, from health to homeland security, including financial services in partnership with financial institutions.

This morning I woke up to great news that Microsoft is at the forefront of Deep Learning on voice, imagine what this could bring to Anti-Money Laundering or Insider Trading products.  The other news was that some top quants of Two Sigma just solved an MRI algo to predict heart disease, and I hope other great minds will, as most of them usually do, also give back to society by applying their amazing knowledge to such grand challenges.