How banks can slow OpenAI

EVIDENT AI PATENT TRACKER

PAT SIGNAL

Patents have long been seen as a performative thing. Most of the banks we track in the Evident AI Index feel they’re not worth the trouble. But for the banks that do file them, patents might end up being a useful tool as AI labs move onto their turf.

That threat is growing: Just this week, OpenAI struck a deal with Intuit so users can get personalized credit product recommendations or tax answers inside ChatGPT. As OpenAI and Anthropic push to make LLMs the new financial hubs, banks need to protect what makes their customers loyal in any way possible.

Innovation is one way to do it. The Evident AI Patent Tracker shows banks are increasingly patenting the ways they’re improving their customers’ experiences – the part of their business vulnerable to disruption by the LLMs. Patents alone won’t stop the frontier labs. But as banks race to build their own end-to-end tools, those protections may help them hold onto what their customers value before “banking” becomes something people do through someone else’s chatbot.

Bank of America, for example, has filed more than 25 AI patents that now underpin its CashPro platform, which automates treasury workflows (see: “Multitooling,” The Brief, Aug. 7). More than 70% of corporate clients now use CashPro, the bank said last month. Those patents make the platform a little bit harder to imitate – a way to protect that customer buy-in when the next hot product hits the market.

Others are mounting similar defenses for systems that have potential to keep customers happier. Wells Fargo patented a way to gauge sentiment in conversations and generate personalized responses. TD Bank says it patented features of its virtual assistant and is scaling it across the business. The bank says it specifically improves the way teams work with customers (see more on that in Use Case Corner below).

USE CASE CORNER

IP IN ACTION

TD Bank’s new wealth management virtual assistant is an example of a patent turned into an application. What the tool does isn’t unique. How it was built is.

Use Case: Wealth Virtual Assistant
Vendor: n/a
Bank: TD Bank

Why it’s interesting: The bank is reusing its tech to launch generative AI assistants across multiple lines of business and cutting down the time it takes to do so. This assistant began as a contact center tool, became a tool for the bank's securities arm in July and now hits wealth management. The bank plans to create similar assistants for seven teams by year’s end. 

How they did it: The idea came through the TD Invent platform, which works like a suggestion box. Layer 6, TD’s research arm, then built an application to match what employees said they needed and patented some of its features. Now, it’s taking the core of what the chatbot does – turning employees’ written questions into code that searches internal documents and data – and scaling it to different teams. The bank credits its cloud architecture for the quick rollout – something Capital One echoed this week in a blog about its assistant, Eno.

By the numbers: The tool will improve the efficiency of its wealth team and “ deepen client relationships,” Luke Gee, chief analytics and AI officer, said. It has yet to share definitive ROI numbers.

Differentiating factors: Wealth management AI is crowded: We covered Citi’s wealth assistant back in July (see: “Wealth Helper,” The Brief, Sept. 4). HSBC rolled one out in September. JPMorganChase debuted an agentic tool in May. TD’s assistant stands out not because of what it does, but how the bank did it – collapsing the time to production by reusing its tech. Prism, its in-house foundation model powering it, is one reason for that. Because it’s trained with the bank’s data, it requires less tuning and is quicker to plug into new use cases. It also makes it up to 30% more effective than other models at identifying customer needs, the bank says.

STAT OF THE WEEK

The share of financial services firms mentioning quantifiable benefits from AI during earnings calls and conferences through Q3, according to Morgan Stanley. The sector beat out the S&P500 on the whole, which clocks in at 15%, but trails the information technology (39%) and communications services (26%) sectors. 

Bigger picture: The share of the 50 banks we track reporting the total financial benefit of AI (like DBS again did last week) roughly matches the bank’s findings. But on the use case-level, banks show a much deeper benefit: 32 of 50 firms now report use cases with outcomes like revenue uplifts, efficiency gains or customer satisfaction boosts.

IN THE NEWS

NEW WEEK, NEW MODEL

Google released Gemini 3 on Tuesday, which, according to CEO Sundar Pichai, can “get what you need with less prompting.” The model beats OpenAI and Anthropic on a number of benchmarks, and early reviews are positive, with Ethan Mollick saying it “demonstrates the change from chatbot to agent.” But evaluations inside banks will be the true test of whether it’s the right fit for use cases (see: “On-paper tiger,” The Brief, Nov. 13).

Bank regulatory teams’ workload may get lighter: The stymied moratorium on state-level AI regulation in the U.S. could come back as an executive order. The leaked draft would create an “AI Litigation Task Force” to undo the “patchwork regulatory framework” in favor of a “minimally burdensome national standard.” It’s not the only way the issue is being pushed: This week, House Majority Leader Steve Scalise said it could also be added to the National Defense Authorization Act, scheduled for a vote next month.

Bank of America’s AI assistant Erica can now handle the equivalent workload of 11,000 customer service reps, the bank said. It’s a showcase of how deep AI is getting embedded, but a reminder that everything banks automate needs a backup strategy. “If people lose trust in that answer [from Erica], 11,000 people have to be put on the phones and in the branches tomorrow. Tomorrow,” said CTO Hari Gopalkrishnan. 

Santander and MUFG invested in Sakana AI’s $135 million series B this week, which made the lab Japan’s most valuable unicorn. The firm published research this summer showing how to get LLMs to work together to decide which model is right for a particular task, a model agnosticism boon (see: “How to hook your banker on AI, The Brief, July 10).

NOTABLY QUOTABLE

TALENT MATTERS

C-SUITE SKILLS

Scotiabank elevated Phil Thomas to chief strategy and operating officer. Thomas was the bank’s chief risk officer and will bring “a deep focus on elevating the client experience and AI innovation to his role,” the bank said.

Deutsche Bank hired Vanessa Yiu as group technology infrastructure CTO. Yiu was previously UBS’ head of core engineering and was a managing director at Goldman Sachs before that.

Yogendra Gadilkar joined UBS as a director where he’ll build generative AI tools, he wrote on LinkedIn. Gadilkar previously worked on research automation tools at Citi. 

Citizens Bank is hiring four Gen AI roles, including a data and AI product strategy leader.

CODA

AT YOUR SERVICE

After a long career break, Jeeves (of Ask Jeeves fame) has a new job at Deutsche Bank.

During the German lender’s investor day this week, private bank chief Claudio de Sanctis said a “banking butler” was on its way. It’s a “voice-enabled” tool customers can speak to through its app to sort out their issues. The butler will eventually provide “advisory and investment solutions.” And later on it will become the tool for “non-stop banking.” 

While one German bank puts a voice to AI, another is putting an actual face. Commerzbank created an avatar named Ava this year to help customers “get assistance with their banking transactions.” Research shows people trust chatbots more when they have human traits, which begs the question of what the future of banking looks like – talking to a human-like bot on your phone? Chatting to a hologram? Or something stranger?

In the immediate future though, at least at Deutsche Bank, the future looks a lot like asking Jeeves.

Tues 25  - Weds 26 Nov
AI for Finance, Paris

Sun 30 Nov - Sun 7 Dec
NeurIPS, Mexico City & San Diego

Tues 2 - Thurs 4 Dec
Global Banking Summit, London