Reading the bones: Citi, BNY, Morgan Stanley invest in AI, alt data, & private markets

If you wanted an idea of what the future looked like in 3300 BC, you’d go to a shaman. Within Inner Mongolia, radiocarbon analysis of shoulder blades found at dig sites revealed that the Mongolians were among the earliest practitioners of pyromantic scapulimancy—a process through which animal bones were washed, put onto a fire, and observed carefully. The results of the process would be compared to a list of outcomes in a companion manual, and the diviner’s reputations would be determined by their predictions’ accuracy.

If you want an idea of what the future looks like in 2024 AD, you go to an investor. Neither method is exact, but modern investors have 5,000 years of experience on their predecessors, and predicting emerging trends and auguring future technological change is not only still a job, but it’s one in which top financial services firms are investing heavily.

Selma Bueno, global head of Morgan Stanley’s Inclusive Ventures Group, the firm’s in-house accelerator for startups, says that what’s popular at present is not necessarily what will be popular in the future.

“I think it’s a really interesting time to invest as venture capital always has a good insight of what’s coming because you’re investing in these companies that are looking at disruptive technologies and identifying really cool opportunities,” she tells WatersTechnology. “It’s what’s going to take off in a few years; it’s not about what is doing well right now. We saw a lot of AI talk maybe a year or two before that became the big thing.”

Morgan Stanley’s Inclusive Ventures Lab partners with a variety of startups across multiple sectors, and leverages the bank’s size and reach, as well as investing capital, to identify interesting use cases and grow specific companies. This year, more than 2,500 startups submitted applications to partner with the firm, but the program has a cap of 25. Bueno notes that while not all of the companies the lab invests with are strategic to the business, many of them are—including its interest and investment in the private markets.

“Right now, we see a lot around data to do with alternative assets,” she says. “How do you measure that? How do you see the impact of alternative funds, because all that data is not totally public, right?”

Industry interest in the private markets jumped last month when investment management firm BlackRock acquired private markets data provider Preqin in a deal worth more than $3 billion. BlackRock CEO Larry Fink said on the company’s most recent earnings call that by integrating Preqin with the investment manager’s flagship order management system, Aladdin, the company can “bring the principles of indexing to the private markets.”

Bueno says potential investors in the private markets want more data before they make decisive moves.

“People are really investing in the technology to try to gather all that information to make investors smarter about how they choose where to invest their money in terms of alternative assets of all sorts,” she says.

Preord(AI)n

Once data has been acquired, it needs to be processed. But this timely process is reduced by leveraging a new technology, currently in the midst of a frenzied hype cycle: generative AI.

One reason for the sustained GenAI enthusiasm since the November 2022 public launch of ChatGPT is that most predictions about the tech were accurate. Chatbots make research decisions easier, even if there are occasional risks of using generative technologies—but these were forecasted beforehand.

Vibhor Rastogi, global head of AI, machine learning, and data investments at Citi Ventures, Citi’s investment arm, says that investing in the technology has become table stakes for financial institutions, but he says that the firm has been “very thoughtful” about its use of AI.

“I personally think that every financial institution has to invest in AI and use AI as a business imperative,” he says. “Just like prior technologies like cloud, mobile, and others, I think financial institutions are seeing that they could be competitively disadvantaged if they don’t use AI, so I think in that sense, yes, it’s becoming table stakes.”

Rastogi has been in the venture capital and private equity space since 2008, where he started at corporate VC firm Intel Capital, before spending a few years at SymphonyAI, an AI-focused private equity fund. He joined Citi in 2021 to lead the firm’s AI practice. Citi Ventures has been around since 2010, and it invests globally in enterprise technology and fintech for the firm.

Rastogi explains that last year the team at Citi solicited more than 200 use cases across the bank for AI and formed a group comprising business heads and Citi’s chief technology officer, David Griffiths, to narrow that number down to 12. One of those pain points was search optimization and data summarization, and Rastogi partnered with Sandeep Arora, head of Citi institutional strategic investing, to find a suitable company.

The company they settled on, Glean, allows for enterprise search capabilities while leveraging retrieval-augmented generation (RAG) technology. Having identified both a use case and a company that could help with that use case, Rastogi needed to understand both the security of the company and the potential roadmap that Glean’s team saw themselves following.

“Not only do we need all the security and technology architecture to meet a large enterprise’s business compliance standards, but we also want to ensure that the startup’s functionality and roadmap is such that we expect them to continue to remain one of the best companies in the space,” he says. Once the checks were complete, Citi Ventures made an investment in Glean’s recent funding round, which saw the company valued at $2.2 billion.

Last year, Citi Ventures invested in Lexion, a legal tech company acquired recently by Docusign, but it also boasts formidable AI products in-house.

Citi’s Policy Assist Tool, a ChatGPT-like interface on OpenAI, is providing answers to policy-related questions for employees within the firm, Rastogi explains. He says that as of late July, 95% of Citi developers have access to GitHub Co-Pilot, where they can experiment with generative AI technology in a controlled environment. Rastogi says that security is crucial to implementing AI across the firm.

“All in all, we’re moving fast while continuing to be thoughtful and deliberate about security because we think GenAI is a transformative technology,” he says.

Alternative rocks

Identifying interesting datasets and then processing them with technology is part and parcel of back-office life at sell- and buy-side firms, but there are strands of data not typically included in the classical canon that are attracting interest. Alternative data—data collected from non-traditional sources—is the darling of hedge funds and asset managers looking to consistently find alpha by introducing and trading on niche datasets that competitors do not necessarily have access to.

Examples of alternative data can be credit card transaction data, which reveals insights into consumer spending habits during periods of economic uncertainty, or foot traffic geolocation data derived from satellites, which can measure the health of certain companies through in-person store visits. In 2020, a notable application of alt data included linking rises in negative reviews of US candlemaker Yankee Candle with new waves of Covid-19, which infamously caused the loss of smell.

However, due to its nature, this data comes unstructured, and it needs to be standardized, cleaned, and sorted before it can be used effectively. BNY, which recently dropped Mellon from the bank’s name, is one of a number of financial services organizations taking an interest in unstructured data. Marianna Lopert-Schaye, global head of strategic partnerships, investment, and innovation (SPIN) at BNY, says that there is “tremendous excitement” about how much unstructured data is out there.

“I think there’s still tremendous excitement around the fact that there’s a lot of unstructured data, which is where large language models can really make a difference in extracting value and insight,” she says. “We as an industry have typically been very good at extracting insights from structured data like numerical time-series, real-time straight, etcetera. We haven’t, typically, as an industry, been as good at extracting information from unstructured data. There’s a plethora of different types of agreements and different types of unstructured data that sits within that beyond even just thinking about what people think of as research and sentiment analysis.”

Lopert-Schaye explains that SPIN has an enterprise mandate and looks at how it can accelerate strategic outcomes at BNY as well as acquire specific capabilities faster and reduce time to market. She says that SPIN is not necessarily an accelerator or incubator, which builds products to commercialize them, but more of an experimental sandbox where BNY can test out new technologies and partner with interesting businesses. BNY is not just messing around, though. Lopert-Schaye explains that this is a move motivated by internal pragmatism.

“I think that [with] most technologies, there is the peak of hype and excitement and anticipation, and then I think there’s a pragmatism around how long does it actually take for adoption to take place in large organizations that have a lot of regulatory obligations to protect their clients,” she says.

Lopert-Schaye says that her team sees a lot of interest in machine learning, AI, and data curation capabilities in the market, and, like Citi Ventures, see the value of AI tools designed to increase operating efficiency. She says that BNY is currently working with a company in the programmatic labeling space, a process that uses scripts to reduce time consumption as well as the need for manual annotation. Lopert-Schaye says the company’s offering could significantly reduce the amount of time that subject matter experts need to spend training computer models which would free them up to do other activities like interact with clients.

“[We] think about this as both transforming the amount of time it takes [to increase] the accuracy [of AI-based models] to support a human in the loop, and then potentially enabling someone to actually spend a lot more time with the client or to explain things than actually having to spend a lot more time traipsing through documents to find information,” she says.

A little help from my friends

Some banks are employing external shamans for the job of identifying potential startup partners. Brent Fierro, a principal at growth equity investment firm FTV Capital, says that FTV execs each work with 15 companies in their specialist area—Fierro’s is fintech—to identify potential issues across individual firms that can be remedied with technology. FTV calls this arrangement a Global Partner Network (GPN).

“I cover capital markets, asset management, and wealth management, and the 15 executives in my fintech GPN sit in those exact seats at global banks and financial institutions, and I’m responsible for checking up with them and making sure I’m current on their key pain points,” Fierro explains.  “For example, I’m sitting with a large regional bank next week to walk through some of their pain points across commodities and derivatives.”

Document summarization and time-saving tools are proving to be popular, Fierro says, as the back-offices at banks seek out ways to reduce the amount of legwork that goes into searching through corpuses of documents.

“When you think about how banks make money and operate, figuring out how to automate back- and middle-office functions can help improve margins and avoid regulatory fines or human error,” Fierro says. FTV recently sold a majority stake and reinvested in Docupace, a Los Angeles-based tech company specializing in workflow and document management automation services for banks.

The relationship between FTV and startups is symbiotic. While not exactly the same kind of relationship as those found in an accelerator or incubator like Morgan Stanley’s Inclusive Ventures Lab or BNY’s SPIN, both parties benefit.

“We’ll bring [startups] in and say, ‘Hey, can you be a relevant partner for us here to help drive the penetration and the relationship with that bank down the road?’” Fierro says. “With commercial partnerships sometimes comes investment. We recently actually did that with JP Morgan, with one of our wealth management businesses.”

The benefit of multiple large institutional firms wanting to understand more about emerging technology is that it leads to a market that is more tailored to the wants of banks and asset managers. But it was not always this way, Morgan Stanley’s Bueno explains.

“You always have to remain humble and understand the dynamics change, the markets change,” she says. “If you’ve been following this for a little while, people were writing checks for companies that were just a PowerPoint presentation not too long ago.”