Vise Intelligence wants to use AI to assist financial advisors

Of all the industries rushing to embrace generative AI, it seems odd that we haven’t heard — nor reported more on fintech, and more specifically, financial advisors.

But here comes Vise to buck the trend. The seven-year-old New York City fintech former unicorn co-founded by two (at the time) 16-year-olds suffered a bout of bad press over the last several years, losing 35% of its assets under management (AUM) in a matter of months, only to see Business Insider reporting on the departure of its largest client, Manhattan West, and the loss of more than 100 employees since its start, through attrition and layoffs.

The Business Insider report suggested the co-founders’ youth, inattention, and inexperience may have led to these issues, and the duo later admitted to RIABiz they needed to change course and do a “hard reset.”

Now, Vise is ready for a big comeback and more focused than ever with the release of its new AI service, Vise Intelligence, a conversational AI model designed to support human financial advisors by preparing them reports, answering their questions, and surfacing up-to-the-minute information about investment portfolios to go over with their clients.

“Supporting financial advisors with artificial intelligence that can help do their jobs better will create better investment outcomes for all the clients that use them, and make [the financial advisors] more accessible to more people,” said Samir Vasavada, CEO and one of the firm’s original co-founders, in a video call interview with VentureBeat.

For example, an financial advisor who pays for a Vise Intelligence subscription could prompt the assistant with the phrase “Ava Harris called with concerns about investing in energy companies,” referring to a client’s phone call. Vise could then provide the financial advisor with information about that client’s specific portfolio, suggest ways for the advisor to tweak to the client’s wishes, and then draft an email to send to the client about the changes the advisor would implement in the investment strategy.

In this way, the human advisor and their client remain in control, but Vise Intelligence is always standing by to act as a helpful assistant capable of pulling together information and suggesting how it could be used.

Where AI and fintech collide

Vasavada co-founded Vise back in 2016 alongside Runik Mehrotra, still its chief investment officer (CIO), a year before the generative AI boom got started with the publication of the “Attention Is All You Need” paper by Google researchers on arxiv that led to the transformer model architecture now underpinning Vise Intelligence and most other leading AI models, such as OpenAI’s ChatGPT.

Previously, the company offered “highly personalized portfolios, fully automating the investment management process, and providing deep insights on each investment decision,” according to one of its earlier funding announcements.

As for what specific AI is being used to enable Vise Intelligence, Vasavada didn’t provide details, but in a Medium post published today, the CEO wrote: “Vise Intelligence is powered by cutting-edge large language models, which we’ve fine-tuned using relevant investment and portfolio management data.”

RIABiz reported earlier that Vise planned to “incorporate new AI models, like ChatGPT, where applicable,” and quoted Mehrota saying it would “end up building functionality that sits on top of one of these pre-trained models … we’re not always going to be internal forever … [what] we build will be technology that sits on top of models these AI companies come out with,” so it is probable that GPT is powering some of the tech.

However, Vasavada did not that Vise Intelligence was designed to ingest information from every specific client in an financial advisor customer’s portfolio — from the “high net-worth to the low-net worth… small-single person firms that are managing money for teachers and firemen, boutique firms with 20 or 30 advisors that are managing hundreds of millions of dollars for executives, enterprise wealth management firms that manage hundreds of billions of dollars in assets for all kinds of different clients,” in his words — and custom tailor its insights for them, so their human advisor could talk them through what was happening with their money.

Leveraging market data and individual client goals

Vasavada also noted that Vise Intelligence was trained on “tens of thousands of data points from the market on different companies, fundamentals of companies, broad market trends” and that this training was combined with information from each client in a secure inference.

“This is previous information and forward looking information,” Vasavada clarified, including the client’s previous positions, trades, gains and losses, as well as their financial goals, risk tolerance, retirement date, other financial milestones such as sending kids to college or purchasing homes.

All this is combined, in turn, with guidance from the financial advisor telling Vise Intelligence what kinds of strategies and investment opportunities the advisor wishes to follow to fulfill their clients’ goals.

“Think about it as inputs from the market, the client and the advisor,” Vasavada said.

Vise did not specify how it secures client data, but Vasavada said “the data is very secure and protected.”

Ultimately, the company believes that “wealth management is going through a transformation,” according to Vasavada, one wherein “rechnology and investment management are no longer going to be separate.” And Vise wants to be the one financial advisors turn to when looking to serve more clients with a human touch.

The ideal scenario for Vise’s financial advisor clients is that “you went from managing 100 clients before to being able to manage 150 clients, and all of your time is being spent on the thing you love doing, which is managing and building client relationships,” the CEO told VentureBeat.

That kind of scaling, of course, also benefits Vise, which takes a cut of its clients earnings from their services and their clients’ portfolio performance.