Despite the dynamism of the artificial intelligence (AI) landscape, startups and tech titans alike are finding their place in this ever-evolving field. Rather than striving for perfect order and predictability, some companies are saying “no thank you” to rigidity—embracing the uncertainty and ambiguity that AI inherently creates.
Brex is among the growing number of companies unapologetically leaning into the “messiness” of AI, discovering it not as a weakness, but as a strategic advantage—and potentially a game-changer.
The Brex Method: Tactical vs. Strategic
Brex provides tools for:
- Corporate expensing
- Budgeting
- Financial planning
…primarily targeting startups, mid-size, and large companies.
Faced with the rise of generative AI and machine learning, Brex reached a crossroads:
- Option 1: Enforce structure and precision across all workflows
- Option 2: Accept that AI—and its underlying data—is far from perfect
Brex chose the latter. Instead of fearing data ambiguity or volatile customer behavior, the company built systems that anticipate and tolerate variability—transforming that into opportunities to iterate faster and deliver better outcomes.
“AI is most powerful when you stop trying to control every variable,” said Brex co-founder and co-CEO Henrique Dubugras.
“At Brex, what we’ve learned is that the sooner we can experiment, learn, course-correct—and be comfortable with that process, messiness and all—the better we’ll be able to deliver value for our customers.”
Messy Data, Real Insights
In financial services, speed is crucial, but data quality is equally important.
- Transactions are structured
- Human behavior around spending is not
- Natural language data (memos, chats, invoices) adds complexity
Brex’s Response:
- Builds ML models that:
- Categorize clean expense records
- Handle incomplete, inconsistent, or missing data
- Trains AI using:
- Transaction histories
- Approval workflows
- Employee communications
Brex’s AI does not require perfect input. It embraces contradictions to deliver more accurate predictions and intuitive user experiences.
Embracing Rapid Iteration
Agility is Brex’s currency in AI development.
- Rebuilt its product development to focus on:
- Frequent, small-batch testing
- Fast shipping of experimental features
Key Infrastructure:
- Proprietary internal tools for:
- Simulating user behavior
- Monitoring model performance in real time
- Rapid course-correction based on live data
“Our mode is changing from ‘get it right before it ships’ to ‘ship it, see what happens, and then get it right,’” shared a product lead.
“The A.I. world doesn’t wait for perfect answers.”
AI as a Co-Pilot, Not a Replacement
Brex positions AI as an assistant, not a replacement. Human context remains critical in situations like:
- Unusual transaction approvals
- Detecting potential policy violations
- Interpreting complex travel expenses
AI Enhancements Include:
- Surfacing suggestions
- Flagging anomalies
- Automating repetitive tasks (e.g., receipt matching, report generation)
Natural Language Processing (NLP):
- Enables users to ask plain-English questions like:
“What’s our marketing spend in Q2 across departments?” - Returns formatted answers, often with visuals
These features reflect Brex’s dedication to human-centered design, even as AI powers the engine.
Creating a Culture That Embraces Ambiguity
Brex’s AI advantage doesn’t stem from tech alone—it’s rooted in culture.
Key Cultural Elements:
- Appreciation of uncertainty
- Encouragement to:
- Challenge assumptions
- Question data quality
- Test hypotheses quickly
Talent Strategy:
- Prioritizes problem-solving over credentials
- Seeks employees who are:
- Comfortable in gray areas
- Skilled at iterating on incomplete data
Ethical AI Practices:
- Teams trained on:
- Model biases
- Edge cases
- Failure scenarios
- Transparency is promoted about what AI can and can’t do
Brex fosters realistic, ongoing discussions to ensure grounded decision-making amid rapid AI advancement.
What’s Next: AI as a Distinctive Long-Term Advantage
As AI reshapes every business, Brex sees messiness as a competitive advantage. While others focus on data cleanliness and tighter controls, Brex:
- Doubles down on flexibility and learning
- Uses AI forecasting tools to:
- Model multiple budget scenarios
- Flag financial risks
- Suggest policy optimizations based on user behavior
These tools are not static—they evolve with each new data point.
Conclusion
Brex’s journey through AI illustrates a compelling truth:
The road to innovation is rarely neat—and that’s okay.
By:
- Building flexible systems
- Choosing speed over perfection
- Keeping humans in the loop
Brex is forging a bold, pragmatic path in the AI era.
In an industry that favors polished demos and ideal results, Brex’s embrace of gritty realism might just be the edge that sets it apart.
