Embracing the AI Revolution: How Brex Redefined Procurement to Keep Pace

The rapid evolution of artificial intelligence has presented a significant challenge for companies accustomed to slower, more deliberate technology adoption cycles. As new AI tools emerge at an unprecedented pace, many enterprises find their traditional procurement processes simply cannot keep up. This mismatch creates a bottleneck, hindering innovation and risking companies being left behind in a rapidly changing competitive landscape.

Corporate credit card and financial management company Brex encountered this exact challenge. Like many of its larger enterprise counterparts, the startup realized its established methods for evaluating and acquiring software were ill-suited for the age of AI. The solution? A fundamental rethinking of their procurement strategy, designed to inject speed, flexibility, and employee empowerment into the process.

The Procurement Predicament in the Age of AI

Traditionally, software procurement in large organizations involves a lengthy, multi-stage process. It often includes detailed requirements gathering, vendor evaluation, security and legal reviews, piloting periods, and extensive internal approvals. This can easily take months, sometimes even over a year, from initial interest to final deployment.

While this deliberate approach offers control and reduces risk for established, stable technologies, it becomes a significant liability when dealing with a field as dynamic as AI. The capabilities of AI models and tools are constantly improving, new startups are launching innovative solutions weekly, and the market landscape shifts rapidly. A tool that seems promising at the start of a months-long evaluation might be surpassed by a competitor, become obsolete, or lose the interest of the team that requested it by the time the process concludes.

James Reggio, CTO at Brex, highlighted this issue at the HumanX AI conference. He explained that their initial attempts to use their standard procurement strategy for AI tools quickly proved ineffective. "In the first year following ChatGPT, when all these new tools were coming on the scene, the process itself of procuring would actually run so long that the teams that were asking to procure a tool lost interest in the tool by the time that we actually got through all of the necessary internal controls," Reggio stated.

This experience served as a wake-up call. Brex recognized that clinging to outdated processes would mean missing out on the potential benefits of AI and falling behind competitors who could adopt new technologies faster.

Brex's Strategic Pivot: Building an Agile AI Procurement Framework

Responding to this challenge, Brex embarked on a mission to completely redesign its approach to AI tool procurement. The goal was to create a system that could move at a pace closer to that of AI innovation itself, without sacrificing necessary security and compliance.

The first step involved streamlining the foundational elements of the procurement process, particularly legal and data processing agreements. Reggio noted that the company developed a new framework specifically tailored for AI tools. This allowed them to vet potential tools more quickly and get them into the hands of employees for testing much faster than before.

This initial acceleration of the legal and compliance review is crucial. Many traditional procurement delays stem from these stages, especially when dealing with novel technologies like AI, which raise new questions about data privacy, security, intellectual property, and bias. By proactively developing a framework that addresses these concerns efficiently for AI applications, Brex removed a major bottleneck.

The 'Superhuman Product-Market-Fit Test': Empowering Employees

Beyond the initial vetting, Brex implemented a novel approach to determine which tools were truly valuable and worth investing in long-term. They call this the "superhuman product-market-fit test." This method shifts the evaluation power significantly towards the end-users – the employees who would actually be using the tools daily.

Instead of relying solely on centralized IT or procurement teams to make decisions based on vendor pitches and feature lists, Brex empowers its employees to test tools and demonstrate their value in real-world workflows. This approach acknowledges that the people doing the work are often best positioned to identify which tools genuinely improve productivity, creativity, or efficiency.

Reggio elaborated on this, saying, "We go deep with the folks who are getting the most value out of the tool to figure out whether it is actually unique enough to retain." This means the evaluation isn't just about checking boxes on a feature list; it's about observing and measuring the tangible impact a tool has on employee performance and business outcomes. The "superhuman" aspect likely refers to the amplified capabilities employees gain when effectively leveraging these AI tools, making their judgment on the tool's value particularly insightful.

Delegating Spending Authority: A Bottom-Up Approach

A key component of this employee-centric evaluation is the delegation of spending authority. Brex provides its engineers with a monthly budget, reportedly $50, that they can use to license software tools from an approved list. This list likely contains tools that have passed the initial, faster legal and data vetting framework.

This might seem like a small amount individually, but collectively, it represents a powerful mechanism for decentralized innovation and discovery. By giving engineers direct control over a portion of the software budget, Brex encourages experimentation and organic adoption based on demonstrated value at the individual or team level.

Reggio explained the rationale: "By delegating that spending authority to the individuals who are going to be leveraging this, they make the optimal decisions for optimizing their workflows." This bottom-up approach contrasts sharply with traditional top-down procurement, where decisions are made centrally and tools are pushed out to employees, sometimes with limited buy-in or understanding of their practical utility.

The delegated spending also provides valuable data. Brex can observe which tools are being licensed most frequently by engineers. This usage data serves as a powerful indicator of which tools are providing real value and warrant consideration for broader licensing deals or integration into standard workflows. It's a data-driven approach to identifying successful AI applications within the company.

Interestingly, Reggio noted that this approach hasn't led to a single tool dominating adoption. "It's actually really interesting and we haven't seen a convergence," he said. "I think that that has also validated the decision to make it easy to try a bunch of different tools, is that we haven't seen everybody just rush in and say, 'I want Cursor.'" This suggests that different teams and individuals find value in a diverse set of tools, further validating the need for a flexible, employee-driven adoption model rather than a one-size-fits-all approach.

Outcomes and the Philosophy of 'Embracing the Messiness'

Brex's new approach has led to a significant increase in the number of AI tools being tested and used within the company. Reggio estimated that they are about two years into this "new era where there's 1,000 AI tools within our company." This sheer volume of experimentation is a direct result of removing the traditional procurement bottlenecks and empowering employees.

Of course, not every tool proves to be a long-term fit. Reggio acknowledged that they have "definitely canceled and not renewed on maybe five to 10 different larger deployments." This is an expected and acceptable outcome of an experimental approach. The cost of trying and discarding some tools is outweighed by the benefit of quickly identifying the ones that provide significant value and avoiding lengthy commitments to tools that might not have panned out under the old system.

The core philosophy underpinning Brex's successful navigation of the AI landscape, according to Reggio, is the need to "embrace the messiness." This means accepting that in a field moving as fast as AI, it's impossible to have a perfect, long-term plan from the outset. Decisions will need to be made quickly, based on imperfect information, and some of those decisions will inevitably lead to dead ends.

"Knowing that you're not going to always make the right decision out of the gate is just like paramount to making sure that you don't get left behind," Reggio emphasized. The risk of making a suboptimal choice on a specific tool is far less significant than the risk of inaction – spending six to nine months evaluating everything meticulously, only to find that the technology or the market has moved on. "You don't know what the world is going to look like nine months from now," he added, highlighting the futility of overly cautious, drawn-out processes in this environment.

Lessons for Other Enterprises Navigating AI Adoption

Brex's experience offers valuable lessons for other companies struggling to integrate AI effectively. The key takeaways extend beyond just procurement and touch upon organizational culture, risk tolerance, and innovation strategy.

Prioritize Speed Over Perfection: In a fast-moving field, the cost of delay often outweighs the cost of a potentially suboptimal early decision. Establish processes that allow for rapid testing and iteration.
Streamline Foundational Processes: Proactively address legal, security, and compliance concerns for new technology categories like AI to remove common bottlenecks. Develop templates and frameworks that accelerate review.
Empower End-Users: The people who will use the tools are best equipped to evaluate their practical value. Create mechanisms for employees to test, champion, and provide feedback on potential solutions.
Delegate and Decentralize: Consider empowering teams or individuals with budgets and autonomy to experiment with tools relevant to their specific workflows. This fosters organic discovery and adoption.
Embrace Experimentation and Failure: Not every tool will be a success. Build a culture that views trying new things, and occasionally discarding them, as a necessary part of the innovation process.
Use Data to Inform Broader Strategy: Monitor tool usage and employee feedback from decentralized experimentation to identify patterns and inform decisions about larger deployments or enterprise-wide solutions.

The traditional enterprise procurement model, built for stability and predictability, is fundamentally challenged by the dynamic nature of AI. Companies that succeed in leveraging AI will likely be those that can adapt their internal processes to match the pace of external innovation. This requires a shift in mindset, moving from a gatekeeper model to one that facilitates and accelerates experimentation.

The Future of Enterprise AI Adoption

As AI technology continues to mature and become more integrated into business operations, the methods for adopting it will also evolve. The current phase, characterized by a proliferation of specialized tools and rapid advancements, necessitates the kind of agile, experimental approach that Brex has adopted.

Looking ahead, companies might see a consolidation of AI capabilities into fewer, more comprehensive platforms. However, even then, the ability to quickly evaluate new features, integrate them into workflows, and measure their impact will remain critical. The lessons learned from embracing messiness and empowering employees during this initial wave of AI tools will likely serve companies well in navigating future technological shifts.

The challenge for enterprises is not just identifying the 'right' AI tools, but building the organizational muscle to continuously discover, evaluate, and integrate valuable technologies at speed. Brex's story is a compelling example of how a company can proactively dismantle internal barriers to innovation and create a system that thrives on experimentation and employee-driven insights, ensuring they don't get left behind in the race to adopt AI.

The success of this approach hinges on a cultural willingness to accept a degree of uncertainty and decentralization. It requires trust in employees to make good decisions when given the right framework and resources. It also necessitates robust monitoring and feedback loops to ensure that experimentation translates into tangible business value and informs strategic technology investments.

Brex's journey highlights that staying competitive in the AI era is less about picking the perfect tool upfront and more about building a resilient, adaptable system for continuous discovery and adoption. By embracing the inherent 'messiness' of rapid innovation, companies can unlock the transformative potential of AI and position themselves for future success.

Subscribe to Our Tech & Career Digest