Generative AI will upend fintech
How? I don't know! Why don't we ask it?
As creative pursuits from web design to writing undergo an incredible upheaval with the appearance of generative AI technologies like DALL-E and GPT-3, it’s fair to wonder what effect these technologies will have in other markets. This post is meant to offer a taste of what could happen in fintech.
Last week I got access to a new AI powered word processor called Lex. Unbelievably, as if I didn’t already devote enough ink on this newsletter to Nathan Baschez, founder and author at Every, Lex was created by none other than… Nathan Baschez.
He wrote the story of how he got the idea and brought it to market in this post. The cliff notes of the story is that he felt like there was room for improvement in online word processing for internet writers, and that adding artificial intelligence to generate writing ideas via GPT-3—a text generating AI from OpenAI—could be fun. He built it as a side project, and released it to the market, and 25,000 people signed up to use it in 24 hours.
My favorite part of the story is that during our market analysis class, which I wrote five posts about, Nathan used word processors as an example market, and at the end of the class, we all knew why.
So, this is the background of this post. I’m writing this post with Lex, and I’m wondering what the fintech equivalent of Lex would be.
Let’s figure it out by using AI to help us. In the rest of the post below, all text written in block quotes with the blue line on the left is generated by AI with no editing other than formatting from me.
In order to see how to do a “Lex of Fintech” we first need to understand the keys to what made Lex’s market introduction a success.
1) There is room for improvement in the user experience of existing products. In the word processor world, Google Docs, and Microsoft Word (and others) weren’t absolutely perfect.
2) Adding generative AI to a word processor is more interesting to writers as a tool than as a replacement. Many other companies were out creating products with the focus of creating whole blog posts or whole articles. Nathan wanted to make something to help writer’s block. Remember the “Jobs To Be Done?” post I wrote? Solving writers block is a job that many writers need done for them, so it’s a brilliant way to position a product against incumbents.
3) Making the experience seamless and intuitive is critical. Other generative AI writing tools were using multiple windows to manage seeding the AI vs producing output. Nathan’s tool just makes use of the existing post that the writer has already been working on as the input, and the writer types “+++” and the AI takes it from there, adding the next 6 or so lines of text.
So with this as a backdrop, is it possible to do the same for some fintech products? I think so.
Lex, how would we do this?
Ok, so the first question to consider is what existing fintech products have room for improvement in the user experience?
One example that comes to mind is online banking. There are a few areas where the user experience could be improved. For example, the process of transferring money between accounts is often needlessly complex. Another example is that it is often difficult to track spending across multiple accounts.
To improve these user experiences, a generative AI like GPT-3 could be added to
suggest money transfers between accounts based on past behavior, or to track spending and offer recommendations for budgeting.
Another example of a fintech product with room for improvement in the user experience is investing. For example, it can be difficult to find information on specific companies and stocks, or to get personalized recommendations for investments.
To improve these user experiences, a generative AI could be used to surface information on companies and stocks, or to make personalized recommendations for investments.
Can you even believe this? I didn’t tell GPT-3 anything about types of fintech or ways that it could be useful. It came up with these suggestions on its own. So it could be that the generative AI that will send fintech rocketing forward already exists, and it is already GPT-3. A way of testing this thesis is to imagine that GPT-3 as a financial advisor:
OK, GPT-3, imagine we’re a couple in our 30s, and we’re about to have our first child. We live in Denver, and we pay $2500/month to rent a one bedroom apartment, and another $800/month in payments for our two cars. To add detail to this picture, let’s say one of us is a paramedic, and the other partner is a mechanical engineer, and our combined income is $9500/month after taxes. We want to get a bigger place for the new baby, but we’re also worried about our jobs in a possible recession.
A financial advisor would suggest that we save up for a down payment on a house, and also create a rainy day fund in case we lose our jobs. They may also suggest we invest in a 529 plan for our child’s future education.
Now, let’s imagine that instead of a financial advisor, we have GPT-3.
GPT-3 would read our situation and then generate a few different options for us.
GPT-3 could suggest we save up for a down payment on a house.
GPT-3 could suggest we create a rainy day fund in case we lose our jobs.
GPT-3 could suggest we invest in a 529 plan for our child’s future education.
GPT-3 could suggest we refinance our cars to lower our monthly payments.
GPT-3 could suggest we sell one of our cars and use the proceeds to pay down debt.
Each of these options has pros and cons, and it would be up to us to decide which option is best for our situation.
However, having GPT-3 as our financial advisor would give us the benefit of having a tool that can generate personalized recommendations based on our specific situation.
So there you have it. Fintech needs a Lex.
The artificial intelligence came up with ideas for me, explained how they would work, and then circled back and closed out my argument.
Surely there are some fintechs already out there today looking at examples exactly like this to figure out how to incorporate GPT-3 into a workflow.
The ones that figure out how to most seamlessly incorporate GPT-3 into a valuable workflow and position their product as doing a job the market needs done will lead the competition.
Thanks for reading! Let me know what you think of generative AI and what it’s going to do to fintech.
—Jon Christensen