3 Fintech News Stories
#1: We Don’t Need Customers. We Just Need Their Credit Cards.
What happened?
An agentic payments company raised some money:
Nekuda, a startup building infrastructure for agentic payments, today announced a $5M funding round led by Madrona Ventures with participation from global financial leaders including Amex Ventures and Visa Ventures. The funding will accelerate Nekuda’s mission to enable AI agents to make secure, autonomous online payments by managing payment credentials and authorizations and setting up security guardrails.
So what?
The concept of agentic commerce — creating intelligent, autonomous software agents leveraging large language models (LLMs) that can conduct commerce transactions without human involvement — has been bugging me, but I haven’t quite been able to put my finger on why.
One obvious concern is trust. Can consumers and business owners trust LLM-powered agents to conduct transactions on their behalf? Can these agents become reliable enough that we feel comfortable letting them do their thing without constantly double-checking their work? And can we create new liability frameworks to govern what happens in situations when an AI agent makes a mistake?
These are important questions, but I think they’re solvable.
The more important question (at least to me) is, why are payments companies and commerce enablement companies so excited about agentic commerce? What future do they see? And how compatible is that future with the best interests of the consumers and small businesses that will be using these agents?
Part of the excitement is that LLMs are exceptionally good at understanding intent from natural language, which makes them a superior interface to search for shoppers, as Martin Balaam, CEO of Pimberly, points out:
“Search as we know it will inevitably give way to intent understanding,” Balaam said. “Instead of 10 blue links, shoppers will get a curated set of product suggestions, comparisons and even proactive deals based on their profile and behavior.”
This makes sense to me. LLMs can help users find what they are looking for more efficiently than traditional search, especially when (as is often the case with shopping) the user doesn’t quite know what they want at the beginning of the journey.
But then why aren’t we just calling this thing “agentic shopping”? Why is it critical that these AI agents now only help us shop, but also make payments and complete transactions autonomously?
The language in this quote from the Nekuda press release is instructive:
“We’re excited to support Nekuda in their mission to address the critical last mile of checkout and payment with trust and efficiency in this new era of agentic payments,” said Matt Sueoka, SVP and Global Head of Amex Ventures.
Ahh, the last mile.
That last mile is really bothersome to companies like AmEx and Visa, both of which get paid when commerce transactions are completed (not just initiated), and both of which are investors in Nekuda.
The public commentary from agentic commerce companies and advocates is always very careful to emphasize the importance of keeping customers in control. However, in their heads, I think a lot of the folks at these companies are thinking something more like:
The reason that e-commerce and social commerce and mobile commerce are annoying is that they still require humans to click on buy buttons, which sometimes those humans don’t do. And that’s bad! Bad humans! It would be much better for everyone if AI could just push that button itself, based on some vague expression of intent from a human somewhere earlier in the process.
This is why agentic commerce is so popular. LLM-powered agentic workflows can remove humans from the process while ensuring that the process still happens. There’s not much money to be made in removing humans from the bill payment and renegotiation process. Or the deposit rate shopping and loan refinance process. But there’s a ton of money to be made by removing humans (but not their credit cards!) from the shopping process.
#2: It Gets Worse
What happened?
Another company doing agentic workflow stuff raised some money:
Berlin-based AI voice agent startup Solda.AI has raised a $4m seed round led by Accel, with participation from US private equity firm AltaIR Capital.
Solda.AI has developed AI voice agents to handle sales calls over the phone, with the aim of helping companies boost efficiency. The startup says its AI agents can manage the full range of conversations with a customer — from the first interaction to closing a deal — with the ability to handle 100 simultaneous phone lines at peak hours.
The agents replicate human voices and conversational flow, a crucial advantage given research shows customers are twice as likely to trust a human voice over a more obviously AI-generated voice. Solda.AI’s agents can also speak in multiple languages, including English, Spanish, Portuguese, French and German.
Founded in 2023 by serial entrepreneur Sergey Shalaev, the startup’s voice agents have so far been integrated into over 20 enterprises’ telesales operations, including consumer and SME-focused tbi bank, hair and beauty platform Treatwell and banking and investment app VividMoney.
The company says it generated $7m in incremental revenue for its clients in 2024, with the aim of increasing that to $30m this year.
So what?
Agentic telesales! We’re going in the wrong direction here!
Like agentic commerce, the potential for problems with agentic telesales hinges on incentives.
If your goal is to efficiently sell as much of your product as you can over the phone, you have a challenge. Customers are more likely to buy stuff from humans, but human salespeople are slow and expensive and often reluctant to do immoral or illegal stuff in order to hit their quotas (that distant sound you hear is Dick Kovacevich yelling, “Preach!”) If only there was a way to make customers feel like they’re talking to a human, but, in reality, to have them talk to a robot that is efficient, endlessly scalable, and heartlessly goal-oriented.
The only saving grace with this one is that fairly soon, customers will have their own AI procurement agents answering these sales calls and screening them with ruthless efficiency.
Then it will be back to the drawing board for everyone.
#3: Acorns Earlier
What happened?
Let’s end on an optimistic note, with a non-agentic AI story.
Acorns made an acquisition:
Savings and investing startup Acorns has acquired EarlyBird, an investment gifting platform for families
Founded in 2019, EarlyBird launched a product that combined financial investing with community. The app allowed families and friends to gift investments to children while preserving memories through a digital time capsule. The investments would become the child’s once they turned 18, and they could use funds for things like paying for college, paying a down payment on a home, or seeding their first business.
Wexler and co-founder Caleb Frankel will join the Acorns team to help build out Acorns Early, the startup’s smart money app for kids. Acorns Early offers a debit card designed for kids and teens to help them develop financial literacy and manage their money. The company launched Acorns Early following its acquisition of GoHenry, a startup focused on providing money management and financial education services to 6- to 18-year-olds.
So what?
I’ve been a fan of EarlyBird for a while, and I’ve always like Acorns, so this is a great match.
I like the idea that Acorns is continuing to work backward, in its pursuit of capturing entire households.
First, start with the adults (Acorns actually offers quite a robust suite of banking and investment products these days). Then add banking for kids (age 6-18), with integrated chores/allowance, saving, and financial literacy (that’s the GoHenry acquisition). And now add custodial investment accounts for toddlers and babies!
Now all Acorns needs to do is build out the other half of its platform (financial management and protection for elderly parents), and it will have built the perfect family banking app for consumers age 30-50.
I also think there’s a lesson that other fintech companies can take away from EarlyBird — often, it’s the non-financial services parts of a fintech product that make it sing.
Lots of banks and fintech companies offer UGMA or 529 accounts for parents looking to save and invest on behalf of their kids. The thing that makes EarlyBird stand out is the integrated digital time capsule, which allows parents and other adults in a child’s life to attach videos, images, and written messages to the investments they are making in that child’s future.
I mean, what other bank or fintech app allows customers to capture memories like this:
Well I just got called in because you punched a boy who said lightsabers were ‘only for boys.’ While we’ll talk about using our words instead of fists, I have to tell your future self how proud I am that at just 6 years old, you’re already standing up against anyone who tries to limit what girls can do.
This is the type of thing fintech, at its best, can do. And I’m glad Acorns is helping to perpetuate it.
2 Reading Recommendations
#1: AI Agents as adversaries for building fraud models (by Ayo Omojola) 📚
Just to prove that I’m not an Eeyore on all agentic AI topics, here’s a post from Ayo that I really enjoyed. AI agents have an obvious value in red teaming financial service providers’ fraud defenses, especially when a new product or feature is shipped.
#2: Consumer financial data and non-horizontal mergers (by Linda Jeng, Jon Frost, Elisabeth Noble, and Chris Brummer) 📚
This is a bit nerdy, but I love the core premise — the lens we use to evaluate the competitive threat posed by mergers and acquisitions doesn’t consider one of the most valuable assets that any company operating in financial services can have: data.
As we update our thinking on M&A and antitrust, we should consider this point very carefully.
1 QUESTION FROM THE FINTECH TAKES NETWORK
There are a TON of interesting questions being asked in the Fintech Takes Network. I’ll share one question, sourced from the Network, each week. However, if you’d like to join the conversation, please apply to join the Fintech Takes Network.
What hot takes do folks have on Chime, now that it has filed its S-1 and is in the process of going public?
If you have any thoughts on this question, reply to this email or DM me in the Fintech Takes Network!
