- Brex revamped procurement to adopt AI tools faster, ditching long approval cycles.
- Engineers get $50/month to test approved tools and guide adoption decisions.
- Brex uses real-world feedback, not guesswork, to scale up successful tools.
- The company has tested 1,000+ AI tools, embracing trial-and-error over perfection.
As artificial intelligence continues to reshape industries, companies are racing to keep up with its rapid evolution. Corporate credit card startup Brex found that its traditional approach to software procurement was too slow to meet the demands of modern AI innovation.
Rather than risk falling behind, Brex chose to rethink its strategy from the ground up and embrace a more flexible, user-driven approach.
Rethinking Procurement for a Faster World
Brex CTO James Reggio shared insights at the HumanX AI conference in March. He explained that when AI tools like ChatGPT began gaining traction, Brex tried to evaluate them using its standard months-long procurement process. But by the time a tool was approved, the teams requesting it had already moved on.
This bottleneck prompted a complete overhaul of the company’s evaluation process. Brex introduced a faster framework for legal and data privacy reviews, which made it easier to get new tools into employees’ hands more quickly.
Empowering Employees to Choose What Works
Brex’s new model gives more decision-making power to the people actually using the tools. The company uses what it calls a “superhuman product-market-fit test,” where teams experimenting with AI software are asked to report on how much value it provides in real-world use.
To support this, Brex offers each engineer a monthly $50 budget to license software from a curated list of approved tools. This hands-on approach allows employees to choose tools that fit their individual workflows, rather than being forced into one-size-fits-all solutions.
Reggio noted that there hasn’t been a single tool that all engineers gravitate toward. This variety highlights the importance of flexibility and personalization in tool selection.
Smarter Scaling Through Real Use Data
Brex’s decentralized approach also makes it easier to decide when a broader software rollout is worth pursuing. By analyzing which tools engineers are actively using and finding value in, the company can make more informed decisions about where to invest in larger licenses.
So far, Brex has tested more than 1,000 AI tools internally and has canceled several deployments that did not meet expectations. Reggio sees this as a necessary part of staying innovative. Some tools fail, but that trial-and-error process leads to better outcomes in the long run.
Embracing Uncertainty as a Competitive Advantage
Reggio believes that the biggest mistake companies can make today is overthinking. He warned that lengthy evaluations can lead to missed opportunities, especially in a market that is changing month by month.
“Knowing that you’re not going to always make the right decision out of the gate is paramount,” he said. “What matters most is acting quickly and learning as you go.”
Brex has chosen to lean into the chaos of the AI boom rather than resist it. By trusting its teams, simplifying bureaucracy, and being open to failure, the company is building a faster, more adaptive path into the future.
Follow TechBSB For More Updates
