Why finance leaders are embracing AI-powered spend management

The hype and FOMO around AI (artificial intelligence) have never been higher. Companies that might be doing something with AI (or have “AI” in their name) are getting huge valuation premiums. The AI Twitter threads are getting ridiculous. Every day there are countless viral threads:

• If you aren’t learning AI then someone who is will replace you

• Here are the 100 new things about AI that you MUST know (and tomorrow there will be 100 more)

• AI is going to kill us all

• AI is going to save us all

Recently I have seen the get-rich-quick FOMO a lot — “20-year-olds are making buckets of money with AI. Learn how you can too!”

It is very reminiscent of the “Have fun staying poor” catchphrase from the crypto folks at the peak of that bubble.

The difference between crypto and AI is that there are tons of immediately valuable use cases with AI.

Innovation from hype

Anthony Pompliano had a great recent post on this topic:

These hype cycles around new technologies, whether AI or Bitcoin, always get a bad name from critics. But the hype cycles are essential in seeing the technology flourish. You need excitement to get people to leave their old jobs and come build products or companies in the new industry. You need excitement for investors to part with their hard-earned money and invest in the new industry. And you need excitement to break through the noise and capture the attention of potential new users.

I might not be a Bitcoin bull (I stayed poor), but I agree with the above sentiment.

While tech bubbles can cause a lot of pain for investors, they can also produce very valuable innovations. I am not going to speculate if we are in an AI bubble (or where we are at in the hype cycle). But a lot of money is being poured into AI by investors. In fact, if you aren’t doing something with AI, then Series B investments and beyond are basically nonexistent.

And generative AI unicorns are regularly being minted. But a lot of generative AI companies that are getting funded will likely go out of business. However, given the excitement and level of investments in AI, some amazing world-changing innovations that otherwise wouldn’t have happened will come out of the “AI bubble.”

Who wins the AI race?

With most hype cycles though there are a few major winners and a ton of losers.

In regards to the finance tech stack (probably similar to many other sectors), I would guess that most value accumulates to either 1) the large incumbents who have lots of proprietary data and/or 2) companies doing truly innovative things with AI that others are afraid to do. The latter is where startups will always have the advantage. Big companies with lots of customers are usually afraid to roll out new technology like AI due to:

1. Business model conflicts

2. Technology concerns/gaps

3. Concerns with things like legal, privacy, data, etc.

Startups have the luxury of going fast and blowing things up.

What is Brex doing with AI?

The No. 2 org on CNBC’s disruptor’s list was Brex (a unified spend management tool) and is right behind the company that started the recent AI craze, OpenAI.

Brex must be doing interesting stuff with AI if they are right behind OpenAI on this list, right? So I talked with the Brex team and got details from Brex co-founder and co-CEO Henrique Dubugras.

What is Brex?

Brex is a spend management platform that combines global corporate cards, expense management, reimbursements, travel, and bill pay to help companies control and analyze global spend.

How is Brex leveraging AI?

While mainstream AI is new to a lot of folks, Brex has been at the forefront of using AI to innovate finance for some time, including building tools to:

• Automate compliance and mundane accounting tasks

• Adding receipts, matching transactions, creating transaction memos, categorization of expenses, etc.

• Detect fraudulent card and cash transactions, applications, and customers

• Using proprietary fraud models to help catch fraud

• Automatically underwrite new customers and adjust limits for existing customers based on risk

Brex also recently announced AI-powered tools for finance professionals, one of which is much like an AI chatbot for CFOs to get real-time insights related to spend trends.

How Brex is thinking about AI

Some insights from Dubugras on how Brex approaches artificial intelligence and what makes AI so valuable in a financial solution:

How has AI changed Brex's roadmap?

“We’ve been doing AI for years so it’s always been a part of our roadmap. Now, it’s accelerating to open up more use cases. Accuracy is so much better now. It can replace human touches in many areas. Before, the promises of AI were futuristic. There’s so much more that’s real and possible today.”

How do you differentiate with AI?

• Building for people. “We don't think about AI from a ‘feature’ angle, as something to simply add on to our spend management platform. Rather, we approach AI with specific users in mind — the employee, the manager, the finance admin, the controller — to better understand what kind of AI assistance will make their jobs easier.”

• Vertical integration. “We talk a lot about our all-in-one approach to spend, and it is a real differentiation on training AI/ML models. That’s because those models are leveraging global card data as well as T&E spend, procurement spend, stipends, and more to inform spend-management-level automation.”

• Prioritizing transparency. “With spend and money, transparency is so critical. With Brex, it’s always very clear where a suggestion comes from (human, AI-generated, etc.). We allow users to control how much AI is helping them to build trust among teams/employees. So it’s important to be upfront with anyone using our platform.”

What have been some challenges with adding AI to an established platform?

• “We have enterprise customers who are public companies with tens of thousands of users. There’s a level of care that we need to have vs. a startup without customers. There’s more risk to adding these capabilities and you can’t do it without careful planning.”

• “There’s a lot of interest out there around data protection in AI. It’s our duty and legal obligation to protect customer data, so everything we build starts with the customer in mind.”

• “AI can be expensive. So we are designing to be smart and focused. Not every user will need a seat for every single solution or feature. We are working to turn things on only for the right people. This helps organizations save costs because they only consume as much bandwidth as they need.”