In brief
Dominic Venturo, U.S. Bank's chief digital officer, argues that resilience in banking is no longer just about uptime and disaster recovery. It is the ability to adapt quickly to changing customers, markets and technology. Banks get there by redesigning how they build products, how they plan, and how they use customer feedback, so changes ship in months instead of years.
The main tension is between speed and control. Venturo says rapid change works only if risk, compliance, engineering and product people sit together from the start, with success criteria set in advance. The same logic applies to generative AI, which he sees as a way to remove routine work so staff can focus on what sets their bank apart.
Facts
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Dominic Venturo says U.S. Bank grew from about 20-some thousand employees when he joined roughly 26 years ago to about 70,000 employees.
“We had 20-some thousand employees when I joined, and we have 70 ,000 employees now, so a little bit different over the years.”
0:01:55 · Dominic Venturo · Employee growth
Dominic Venturo says U.S. Bank created its Innovation Group about 14 years ago as part of its payments organization.
“I think the big change in my career was probably about 14 years ago when we created the Innovation Group at the bank, and that was part of the payments organization.”
0:02:26 · Dominic Venturo · Innovation Group
Dominic Venturo says U.S. Bank's early innovation group used a three-horizons approach, building prototypes, testing them with customers and then bringing them to market.
“thinking about the three horizons innovation, thinking about always looking up and over the horizon, doing applied research and development to build prototypes, to test them with customers, and then begin to bring them to market”
0:03:10 · Dominic Venturo · Innovation method
Dominic Venturo says he has run U.S. Bank's digital organization for about four and three-quarter years.
“And for the last four and three quarters, I guess, years, I've had the privilege to run what we call the digital organization at U.S. Bank.”
0:03:49 · Dominic Venturo · Digital organization
Dominic Venturo describes 'zombie products' as offerings banks keep alive despite little traction because they have already spent heavily on them.
“Which is why you see a lot of what I'll call zombie products that are around, which is, you know, we spent a lot of money on it and never really got traction”
0:06:44 · Dominic Venturo · Zombie products
Dominic Venturo says U.S. Bank invested in API development platforms and microservices years ago after seeing fintechs and technology firms move to well-documented APIs.
“we decided to invest in API development platforms and also the microservices capabilities within our technology architecture that would allow that to happen”
0:11:18 · Dominic Venturo · API investment
Dominic Venturo says U.S. Bank can take a well-defined idea to a production version within one quarterly planning cycle.
“we can go from this well-defined thing on a whiteboard that we want to do to a version of it that's in production within a quarterly planning cycle, which is pretty impressive.”
0:12:23 · Dominic Venturo · Delivery speed
Dominic Venturo says customer feedback showed that growing use of password managers caused login problems on banking websites, prompting U.S. Bank to redesign how passwords are entered.
“as the advent of people using password managers more often actually caused some problems with the way passwords are entered on banking websites”
0:13:24 · Dominic Venturo · Login redesign
Dominic Venturo says U.S. Bank has A/B testing capability that lets it deliver a change to a subset of customers and measure results before a full launch.
“we also have built A-B testing capability. So we can have a hypothesis, we can build a change, we can deliver it to a subset of customers and see if it actually performs better before we launch it to the entire organization.”
0:14:01 · Dominic Venturo · A/B testing
Jim Marous says WeBank, which he visited in Shenzhen, China, used four parallel cloud platforms to test ideas and innovated on a 14-day cycle.
“they had four parallel cloud platforms that allowed them to test new ideas instantaneously. They innovated on a 14-day cycle, very close to the time period you're looking at”
0:16:55 · Jim Marous · WeBank innovation
Dominic Venturo says U.S. Bank's digital account-opening team cut the data fields from about 16 to zero, then added back only mission-critical ones, ending with a handful.
“we're going to go from whatever it was, 16 to zero. And then each one that we add has to be mission critical to the objective. And we ended up with just a handful.”
0:20:40 · Dominic Venturo · Account opening
Dominic Venturo says U.S. Bank can now open a new account with two fields of information, with help from technology partners.
“Two fields of info, we can do a new account. Right. With some help from some technology partners.”
0:21:54 · Dominic Venturo · Account opening
Dominic Venturo says U.S. Bank built a banker-facing interface that is essentially the same as the customer interface, letting it reuse code and avoid maintaining separate systems.
“to build an interface for what we call the banker experience that is basically the same interface that the customer uses.”
0:22:25 · Dominic Venturo · Banker experience
Dominic Venturo says many smaller banks rely on core processing or card platform vendors for new capabilities.
“And I know that a lot of, especially the smaller banks, rely on either a core processing vendor or a card platform vendor and the like.”
0:25:52 · Dominic Venturo · Bank vendors
Dominic Venturo says the failure of SVB and other banks produced a deposit crisis that forced banks to make attracting deposits easier and bundle products differently.
“we saw as an industry a pretty significant change when we had the failure of SVB and the other banks. We had sort of the crisis of deposits, for lack of a better description.”
0:29:22 · Dominic Venturo · SVB failure · 2023
Dominic Venturo says U.S. Bank has spent about six years modernizing its architecture and systems, including moving digital applications to cloud-native designs.
“which we've been on that journey for the last six or so years, is modernizing the architecture, modernizing the systems where we can, moving to cloud-native applications in the digital space”
0:30:19 · Dominic Venturo · Modernization
Dominic Venturo says launching a product at U.S. Bank used to take about 18 months because of sequential financial planning, product planning and engineering.
“Because you do your financial planning, then you do your product planning, then you do your engineering, then it would take 18 months.”
0:31:22 · Dominic Venturo · Product timelines
Jim Marous says generative AI did not emerge publicly until November 2022.
“it is amazing in my mind to think that Genre of AI didn't raise its head until November 2022.”
0:31:59 · Jim Marous · Generative AI · 2022
Dominic Venturo says generative AI's breakthrough was enabled by cloud computing and improvements in network infrastructure.
“Part of that was enabled by cloud. Part of that was enabled by network infrastructure improvements and the speed at which we can communicate between where we are and where the cloud is.”
0:33:23 · Dominic Venturo · Generative AI
Dominic Venturo says earlier machine learning models at U.S. Bank were used for image analytics, check processing, fraud, and detecting networks of criminal actors.
“They allowed us to do things like advanced image analytics and check processing and fraud. They allowed us to do things like network detection, not the Internet, but networks of actors in things like fraud prevention”
0:33:57 · Dominic Venturo · Machine learning
Dominic Venturo describes starting with an open-source large language model and training it in the bank's own environment on a knowledge base of policies and documents.
“You start with an open source, a large language model, and then you train it in your environment on a knowledge base, a set of information.”
0:35:00 · Dominic Venturo · Training AI models
Dominic Venturo says banks will probably not build their own large language models because of the money, computing and power required.
“banks probably aren't going to build a large language model because that takes an enormous amount of money and compute and power and the like.”
0:36:41 · Dominic Venturo · Large language models
Dominic Venturo says one-to-one personalization has been hard because bank data is always dated, such as last month's transactions.
“a lot of the reason that that has been very difficult to do is because the data is always dated.”
0:37:04 · Dominic Venturo · Personalization
Dominic Venturo describes unstructured data as information not held in a database, such as financial statements and tax returns, which generative AI can read and extract.
“unstructured data, if you're not familiar, is just not a database. So financial statements, tax returns, people put them in different forms.”
0:37:37 · Dominic Venturo · Unstructured data
Jim Marous says U.S. Bank ranks about fifth among U.S. banks by asset size.
“you're at, I think, number five as far as size of assets.”
0:39:36 · Jim Marous · Bank size
Dominic Venturo says U.S. Bank uses co-browsing and does not force customers into a particular channel.
“We do co-browse and we don't force people into a channel.”
0:40:50 · Dominic Venturo · Customer channels