Users of Credit Karma, the personal finance site
"Our vision at Credit Karma has been to use data and AI to automate as much of your financial life as we possibly can," said Ryan Graciano, chief technology officer and co-founder at Credit Karma.
Banks keep
"With the proliferation of banking as a service, anyone can offer a bank account," said Dylan Lerner, senior analyst for digital banking at Javelin Strategy & Research. "What are you doing to differentiate yourself?"
He finds the strongest way to stand out is with financial health or money management tools — the space where Credit Karma started. Banks will often ask Javelin what nonbank players are getting up to in the financial services space. Lerner finds that nonbank and fintech companies tend to rank last in every category on the Javelin scorecards, including customer service and money movement, except for one: financial fitness. There, they hover near the top.
Banks know their customers are turning to other finance apps, both anecdotally and through aggregation and
This overhaul is also vital for Credit Karma, in Lerner's view.
"Intuit fumbled with the
Intuit is shutting down one of the most popular personal financial management sites of its time — which attracted millions of users despite its counterintuitive approach to accessing consumer data.
Before and after
The older version of the Credit Karma app "felt more fragmented," said Graciano. "It didn't have the places where we could integrate AI easily."
The new dashboard presents an overview of a user's net worth, credit, debt and loan positions when they click between tabs. Before, credit was front and center. Pinning a net worth section to the top appeals to prime customers but also to "anyone who's in their mid-20s and starting to think about what their long-term trajectory might look like," said Graciano. A menu running along the bottom of the screen lets people deep-dive into their credit cards, loans, insurance and Credit Karma Money account.
While the overall design has been pushed out to all users, many modules, explanatory articles created by AI and Inuit Assist pop-ups are still in beta testing.
The layout will be personalized to each user; for instance, someone who is deep in debt may not see a section recommending how they earn more credit card rewards. The Intuit Assist will also materialize at key moments to answer questions (for instance, "which of my transactions are recurring?") or provide context or more information.
The icon is "essentially saying, 'AI saw something, and if you want to talk about it, push this button,'" said Graciano.
The role of generative AI
Credit Karma's efforts demonstrate the opportunities with AI for generating custom editorial content for users based on their specific financial situation. For example, clicking on the button hovering around the credit cards screen may summon an AI-generated comparison chart that breaks down different credit cards especially well suited to this user by type of rewards, rewards rate, welcome bonus, annual fee and specifics around card benefits.
In Graciano's view, this is one way that AI supersedes traditional software.
"I've written software that has compared credit cards in the past and it's difficult because it's such a multi-dimensional problem," he said. "It's incredibly hard to generate this amount of detail in terms of what people would actually want, and that's when they start Googling and piecing it together themselves."
In other cases, Intuit Assist will generate an explanation about why a specific card pick is best suited for the user, or pop up with a spotlight of the best rewards categories in the user's arsenal of credit cards.
"Personalization is the crux of the banking industry right now," said Lerner. "Generative AI is supposed to be the answer to that broad, unruly question — how do I make something for each individual customer of my bank, fintech or credit union?"
Credit Karma typically
Graciano says the company is incentivized to recommend products that suit its users.
"If we can find a financial product that fits you, your satisfaction and your retention go up," he said. "If we find you something you don't like or you can't get approved for, you are much more likely to leave the platform."
These features will work best if users link their financial accounts to Credit Karma — especially their primary spending account and credit cards — but such linking is not required. Credit Karma will also pull data from credit reports and publicly available information about property values of homes and cars.
"AI is really good at processing documents and looking at structured and unstructured data," said Graciano.
The project of revamping Credit Karma's app coalesced about a year ago, with the bulk of development heating up in the past nine months.
Credit Karma's research team interviewed members to find out what they wanted and what they expected when interacting with AI. The project was large enough that the company reshuffled employees and provided cross-training, rather than hiring.
"For most of our data engineers and data scientists, their imaginations have been captured by this for a while," said Graciano.
One key for Credit Karma will be convincing users that these AI-powered tools exist and they should use them, said Lerner. It's unclear to him to what degree consumers understand AI, or are nervous about using it.
Graciano compares it to the release of another product that transformed consumer technology.
"It's like iPhone 2007," said Graciano. "I don't think anyone knew how transformative that would be when it came out."
