When AI tools started exploding onto the scene, most companies weren’t ready.
Many found themselves stuck in slow, outdated processes that just couldn’t keep up with how fast the technology was changing.
Brex, the fintech company known for its corporate credit cards, ran into that same problem. But instead of trying to force AI into their old ways of working, they did something smarter: they changed how they worked altogether.
The Problem: Slow Process, Fast Technology
Brex CTO James Reggio shared their story at the HumanX AI Conference earlier this year.
He explained that when tools like ChatGPT first arrived, Brex tried to evaluate them the usual way—by going through months of internal checks, pilots, and approvals.
But that approach didn’t hold up.
“By the time we were done evaluating the tool,” Reggio said, “the team that wanted it didn’t even care anymore.”
They needed a new approach—one that moved at the same pace as the tech they were trying to use.
The Solution: A New, Faster Way to Adopt AI Tools
Brex didn’t just tweak their process. They completely rebuilt it.
Here’s how:
1. Simplified Legal & Data Review
They created a faster framework to review data policies and legal terms. This cut down delays that used to hold things up for weeks or months.
2. Letting Engineers Lead the Way
Brex gave their engineers $50 per month to spend on AI tools from a pre-approved list. That way, the people who actually use the tools get to decide what’s useful—not just the IT or procurement teams.
“When you let the users choose, they naturally pick what works best,” Reggio said.
What Worked—and What Didn’t
This new approach gave Brex a much clearer view of what tools were actually helping.
If a tool proved valuable, the company could upgrade to a broader license. If it didn’t? They dropped it—no drama.
Reggio says Brex has canceled or skipped renewing 5 to 10 major AI tools in the last two years. They’re not afraid to experiment, but they’re also quick to move on if something’s not worth the investment.
The “Superhuman Fit Test”
To help guide decisions, Brex developed what they call a “superhuman product-market-fit test.”
Here’s what that means:
- They look at who’s using the tool regularly.
- They check how much impact it’s having on real workflows.
- They talk to those users to see if it’s helping them go faster or do better work.
If the answer is yes, they keep it. If not, they move on.
A Key Lesson: Don’t Overthink It
Reggio’s biggest advice for other companies?
“You’re not always going to get it right. And that’s okay.”
He warned that waiting six to nine months to approve tools might mean you miss the moment entirely.
“The biggest mistake is overthinking everything while the world moves on without you.”
Instead, Brex chose to accept a bit of chaos—and it’s paid off.
What Other Companies Can Learn from Brex
Here’s what any business can take away from Brex’s AI journey:
- Speed matters. Move quickly or risk falling behind.
- Empower your team. Let the people who use the tools help choose them.
- Test and learn. It’s okay to try things and cancel what doesn’t work.
- Stay flexible. What works today might not tomorrow—and that’s normal.
Final Thoughts
AI is evolving fast. Brex shows that the best way to keep up isn’t to control everything—it’s to embrace the mess, give your team room to explore, and learn as you go.
It’s not about getting it perfect. It’s about staying in the game.