Klarna Goes All In On AI & OpenAI's Rumored "Strawberry" Model
Here's what's going on in AI x Banking this week.
Hey everyone!
Was a great trip to Charlotte last week for the Datos Financial Crime and Cybersecurity Symposium. We got to meet some amazing, tech-forward, banks that are thinking aggressively about the future of tech and financial services.
AI and synthetic identities were the hot topics—everyone on the fraud and cybersecurity side is worried about identity hacking and deepfakes impacting fraud. Meanwhile, on the AI side, most folks are thinking through the operational efficiencies and benefits as AI moves on from the hype cycle into real use cases.
We also hosted a panel on Scaling Compliance with Automation. Our CEO Will Lawrence was joined by executives from Ally Bank, Truist, Worldpay, and Hawk to discuss the impact Gen AI is having on handling fast growth and doing more with less on compliance.
We had a blast being there, learned a lot, and was a great kickoff to conference season.
As usual a lot going on this week—Klarna seems to be tripling down on their AI bets, and banks are testing new AI products too. Plus, some developments on the foundational model side too.
Let’s dive in.
AI Roundup: OpenAI’s “Strawberry” Model, Banks Using Meta’s LLM
OpenAI’s upcoming model helps with reasoning, not just research
Last week, rumors of a new secret model from OpenAI, called “Strawberry” leaked to the press. The model is said to not just generate answers and text, but also incorporate reasoning, something that’s been elusive for most AI models up until now.
Instead of just asking for simple information like “Tell me about Elon Musk’s life story,” the new model can handle complex math and science questions that require multiple steps. Current models have a lot of trouble with basic logic, even things like Tic-Tac-Toe. The reason the model’s named “Strawberry” is because ChatGPT famously can’t tell how many R’s are in the word.
Reasoning could make a huge impact on the AI ecosystem; rumor has it that the company is getting ready to release it in the fall, so stay tuned on updates.
Meta’s Llama Model Being Used by Banks Like Goldman Sachs
Meta’s open source model, Llama, grew a ton in popularity with developers and companies over the past year. Meta says the model is now being used by banks like Goldman Sachs and Japanese firm Nomura, as well as companies like AT&T.
It’s a big deal because that means banks are not just becoming more open to using open-source tech, but implementing open-source AI at a production level. For competitors like OpenAI & Anthropic, an open-source competitor from well-financed Meta could dramatically affect their bottom line and their enterprise sales strategy.
OpenAI & Anthropic were previously courting banks, financial services, consulting firms and others around essentially becoming dev shops for them, helping them create an AI strategy and build out a product. If banks—which have a lot more engineers than you’d think—can use open-source models that pass their diligence checks to execute these projects in-house, it’ll be a lot harder to sell them on multi-million dollar projects.
Klarna May Cut Its Workforce In Half For AI Products
Some really surprising news out of Klarna this week, which is quickly becoming the leader in the AI & fintech ecosystem.
The company’s CEO, Sebastian Siemiatkowski, said in an interview to the Financial Times that the company may cut 50% of its overall workforce as AI becomes more prevalent throughout the business.
“Not only can we do more with less, but we can do much more with less. Internally, we speak directionally about 2,000 [employees]. We don’t want to put a specific deadline on that.”
Nonetheless, it’s a sign that AI is actually providing operational efficiency for the lender, a great sign for banks and fintechs looking to do the same. The overall business numbers seem to back this thesis up too—the business is a lot stronger now than it was before. In the first half of 2024, the company posted an adjusted profit of $66 million, as opposed to an adjusted loss of $45 million over the same time period in 2023.
It doesn’t seem like this is going to happen overnight, but the fact that the company is even publicly discussing this is eye-opening. The most immediate impact seems to be on hiring—the company is pausing hiring across the board for everything except engineering.
This all also could be a way to generate more buzz and interest around a potential IPO—the buy now, pay later giant was valued at $6.7 billion in 2022 and rumors say that it’s been looking at a 2025 IPO in the US. Bankers say the company could go public at $15 billion to $20 billion.
In any case, two things are becoming clear: that Klarna is becoming a leader in this space and that there’s real value to unlock with AI in financial services.
Wealth Management Tech Could See A $6 Billion Boost From AI
A new report from Technavio says that there could be a big boost to the wealth management space from new tech like AI over the next few years.
The report predicts a 14% CAGR growth over the next 4 years, mainly led from things like blockchain tech and AI.
While the uses of blockchain tech in wealth management is a bit of a head-scratcher, AI makes a lot of sense. There’s a lot of interest around creating highly personalized wealth management strategies using Gen AI, as well as having LLM’s even create and execute trading strategies.
Automating wealth management has been around since the early 2010’s with companies like Betterment, but there’s a chance that Gen AI could take things to the next level. It seems like research firms and incumbents are closely watching the space to see how it evolves.