April 11

BloombergGPT demonstrates how organizations with vast troves of data can create domain-specific generative AI products that can perform better than more general large language models. 

Bloomberg created a massive AI model that could – as one fintech expert put it – “replace the analyst.” 

The company released a research paper that detailed how it trained a new large language model (LLM) on what it describes as the “largest domain-specific dataset” out there. Its training data set included 700 billion “tokens” (read: word fragments), 363 billion of which were taken from Bloomberg’s own financial data (which includes Bloomberg Businessweek stories, TV transcripts, SEC filings, and more). It was augmented with a public data set. 

“We see tremendous value in having developed the first LLM focused on the financial domain,” said Bloomberg’s chief technology officer, Shawn Edwards. The team published test results that showed how BloombergGPT out-performed other general language models in answering queries related to finance.  

While the firm hasn’t launched any specific apps or products yet, BloombergGPT could eventually allow Terminal clients to quickly retrieve information about public companies or ask complicated, multi-part financial-analysis questions. It could also help Bloomberg journalists write headlines.  

As the generative AI rush continues in full force, financial firms are eager to find ways to disrupt their own processes and offerings.  

“Bloomberg is a great example of a domain-based model where an incumbent has just as much opportunity as a startup, because they can use their unique access to data or domain knowledge to codify their IP into these kinds of new intelligent models,” according to Insight Partners managing director Ganesh Bell. “Anyone who has access to that kind of data can do that.” 

For example, Morgan Stanley is giving its financial advisors access to ChatGPT to synthesize its proprietary research, Swedish investment firm EQT programmed a chatbot to help its dealmakers benefit from its “Motherbrain” data platform, Goldman Sachs is experimenting with ways to use generative AI to write code more efficiently, and fintech Klarna has deployed it externally for shopping recommendations.  

Through its Global Markets Incubator, SocGen has honed a way to fast-track innovation, build relationships with cutting-edge startups, and take a leading role in sustainable finance. 

SocGen has a clever way to tap into the buzzy ESG technology industry.  

It just selected eleven startups to join its Global Markets Incubator (GMI), where they will have six months to test and deploy their products within SocGen’s business.  

This year’s cohort includes firms focused on financial carbon metrics, impact investing, biodiversity, and more. The “win-win relationship” gives entrepreneurs access to SocGen experts and feedback, while the bank benefits from “rapid progress and the deployment of innovative solutions.”  

The incubator – which SocGen originally launched in 2018 – ultimately allows it to test tools that it can eventually offer to its clients or use for its own corporate responsibility goals, as environmental, social, and governance (ESG) concerns continue to be a hot topic for regulators, board members, and customers alike. 

“Whether the challenge is climate change, positive impact or governance-related matters, the GMI’s program offers a very exciting opportunity to work with top-of-the-class companies and develop out-of-the-box thinking,” according to the bank’s head of GMI, Antoine Connault.  

Since 2018, more than 50% of cohort companies have entered a commercial relationship with SocGen, Connault told Insights Distilled. For example, Entelligent’s climate science and AI for Alpha’s deep learning techniques are embedded into the bank’s quantitative strategies, while Opensee helps it efficiently visualize and manage its risk metrics, and Nephelai improves its trade booking processes.  

GMI allows SG to “constantly benefit from frontrunning technology innovation, combine our expertise with that of founders and innovation-led companies, and feed our buy/build strategy,” Connault told Distilled. Chosen companies, meanwhile, get to “tap into our expertise, network, reach” in a partnership “that is much more than a usual vendor/client relationship and supports their growth.” 

Here’s the full list of participating companies in the latest cohort: Aspiration, Emmi, BeZero Carbon, Enmacc, Regrow, Net Purpose, Simpl, Greenscope, YvesBlue, Arboretica, and allcolibri. 

This kind of hyper-focused incubator program gives financial firms like SocGen a way to tap into emerging technologies in a deliberate and time-boxed way, speeding up the partnership process. 

 Similarly, Banco Santander recently teamed up with Microsoft on a contest for startups that are using AI to have a positive impact.  

For other techniques that FinServs can deploy to build better relationships with ScaleUps, read our report here.  

Financial institutions have increased their desire for technology that provides real-time data analysis, stress-test scenarios, and communication aggregation as they seek to protect themselves from risk.

This year’s banking crisis continues to have ripple effects for banks’ technology strategies.  

After SVB’s collapse, enterprises are looking critically at their own risk management and planning to up their investments: An analyst predicts in American Banker that “banks will increase their annual spends on risk technology by 8% to 12% in the next 12 months.”  

Financial institutions have long had robust tech teams and robust risk teams, but the events of the past month underscored their need to coordinate more closely in real-time.  

For example, Vanguard CIO Nitin Tandon recently spoke out on the topic of how FinServs’ risk and IT divisions need to come together: “Technology plays a key role in being able to analyze the data, come back with the risk profiles, and develop risk postures.” 

To that point, it’s essential that any data management solution work across organizational silos and take advantage of alternative data sources. For example, the way that communication channels like Twitter and WhatsApp groups contributed to Silicon Valley Bank’s collapse highlighted the need to monitor external datasets. 

Through external partnerships, financial institutions can gain information to help them target SMBs with personalized recommendations on financial content and products. 

A group of financial institutions has linked up with a small business-focused fintech to help them make personalized recommendations. 

JPMorgan Chase, Wells Fargo, and Mastercard are all working with Hello Alice to offer its new Business Health Score tool, which helps SMBs measure and improve their creditworthiness. 

Chase and Wells Fargo will offer the recently announced tool to entrepreneurs and small businesses, while Mastercard will make it available to partner banks that offer small-business credit cards. 

The goal is to simultaneously “empower small business owners” and help FinServs better serve their customers.  

The insights from Business Health Score “provides on a small business owner’s business health and challenges “will benefit financial institutions by allowing them to better identify the needs of small business owners, improve their own engagement strategies and product marketing, and make better recommendations on products and services to SMBs,” Hello Alice CEO Carolyn Rodz told Insights Distilled.  

Hello Alice has reason to believe that SMBs need help: Its research found that only 25% of small business owners have applied for a business credit card, and 85% of those applications were denied due to poor credit or lack of credit. The firm expects that personalized recommendations, education, and products can help turn that tide.  

For example, if a business owner indicates they have a strong revenue stream but struggle with cash flow management, they might receive interactive educational content from Hello Alice alongside information about Wells Fargo’s treasury management solutions.  

The tool is also part of a larger, industry-wide focus on SMBs: Small businesses have historically felt underserved by their financial institutions, leading FinServs to focus lately on hyper-targeted features and products, including Amex’s cashflow-management hub, Standard Chartered’s partnership with upSWOT on personalized tools, and Westpac’s work with Rich Data Co to supercharge its ability to provide credit.  

One of the major hurdles to effective fraud prevention is data disorganization: By consolidating, analyzing, and providing context and connections across previously disparate information, data intelligence platforms can help investigators spot issues, faster. 

A squad of banks just joined the $129 million Series E for Quantexa, a decision intelligence firm that helps break down data silos to combat fraud.  

Most organizations have all the data they need to fight financial crime, but silos and manual processing prevent them from seeing the bigger picture. Master data management tools – which combine information across different source systems – provide transparency and can ultimately help financial institutions understand risk, prevent fraud, and spot opportunities for improvement.  

“Organizations are increasingly in need of a reliable single source of truth within the organization, and traditional approaches and technologies are not solving this need,” chief product officer Dan Higgins told Insights Distilled. Quantexa “deliberately created a truly open architecture that allows for easy integration and industry standard adoption, for everyone from data scientists to analysts,” he added.   

Quantexa’s technology has a wide range of use cases across financial services: It has helped HSBC screen for money laundering across millions of trade finance transactions and launch its Global Social Network Analytics platform, it supported Standard Chartered in investigating criminal activity across billions of data points, and it aided Allianz Insurance in managing risk. 

For example, the firm said it can process 60 billion data records in a single day, with 99% data match accuracy, and a 20% average rate of de-duplication of records. Banking customers have used its platform to reduce financial investigation times from weeks to hours.