Brex reimagines spend management with Claude in Amazon Bedrock

Brex, the modern financial platform combining global corporate cards, spend management, banking and travel, is transforming expense management by integrating AI capabilities directly into their core product offerings with Claude in Amazon Bedrock.

Claude helps Brex to:

• Automate 75% of expense transactions, achieving a 94% compliance rate versus the industry standard of 70%
• Save customers 169,000 hours per month on expenses and accounting through automation
• Save $56.5 million in annual salary costs across all customers
• Review 100% of transactions with AI-driven anomaly detection at the line-item level

Rethinking expense management for AI-first experiences

With a goal of helping businesses spend smarter and move faster, Brex recognized that traditional spend management approaches were reaching their limits. Finance professionals spent countless hours reviewing expenses, while employees wasted time on tedious data entry instead of focusing on their primary work. The status quo required managers and auditors to click through numerous expense reports, often without clear context for why their action was required.

"Every company needs spend management from the moment they have a corporate card or bank account," said David Horn, who leads AI at Brex. "It's just the level of maturity and complexity that changes as they grow. With AI, we can abstract away most of that complexity, allowing both finance teams and employees to focus on what truly matters."

Selecting Claude for enterprise-grade reliability

After exploring various AI models beginning in 2023, Brex chose Claude for its technical capabilities and Anthropic's approach to model development and security.

"What we've valued about Anthropic is not just their powerful models, but how they've positioned them for enterprise needs," said Horn. "When I talk with customers about AI, data privacy is always their first concern—it's the critical foundation we have to address before we can even begin discussing capabilities."

Using Claude in Amazon Bedrock proved advantageous for Brex's enterprise implementation. Jin Shao, a software engineer at Brex, said, "Deploying Claude in Amazon Bedrock gives us seamless integration with our existing Amazon Web Services (AWS) infrastructure and services. Building effective AI applications requires more than just an LLM model—you need knowledge bases, prompt management systems, and robust security controls. Amazon Bedrock provides all these capabilities in one place, reducing our integration overhead."

The AWS environment aligns with Brex's enterprise security requirements, a key consideration with Brex growing its enterprise revenue 80% in the past year and now serving over 200 public companies Shao added, "Since we already host our services and databases on AWS, implementing Claude in Amazon Bedrock was a natural extension that maintained our security posture. This integration allows us to leverage AWS Identify and Access Management (IAM) for access control and keep all data processing within our existing security boundaries, essential for handling sensitive financial information."

Brex also benefits from the operational efficiency of the AWS environment. Shao noted, "The streamlined deployment and monitoring capabilities in Amazon Bedrock have accelerated our development process. We can rapidly test new Claude models as they're released and scale our implementation without managing separate infrastructure, allowing our team to focus on building innovative experiences for our customers rather than managing complex systems."

How Claude powers intelligent expense management

Brex decided to rebuild their platform with AI at its core. Tim Foster, Engineering Manager at Brex, says, "We've spent the past several months fundamentally reimagining the entire spend management stack. Instead of making incremental improvements, we looked at every touchpoint—from the employee submitting expenses, to the manager approving them, to the finance team auditing them, to the overall process optimization. Our central question has been: How can we exponentially improve our customers’ core financial workflows with AI? How can we rebuild each of these experiences from the ground up with AI at the core, rather than simply adding AI features to our existing workflows?"

Claude powers several key capabilities in Brex's reimagined platform:

• Conversational expense submission: Their new spender experience makes expense management nearly invisible to employees. The system analyzes contextual information like calendar events, location data, and spending patterns to automatically categorize expenses and apply compliance rules without requiring manual input. This transforms expense management from a burdensome task into a seamless background process that increases compliance while eliminating administrative work.
• AI-powered policy interpretation: Finance teams can upload policy documents directly into the system instead of complex rule configurations. The AI interprets these documents intelligently, eliminating the need to translate policies into technical rules or decision trees.
• Intelligent expense auditing: An AI-first workbench transforms how finance professionals review expenses by proactively grouping related expenses, flagging policy concerns, and providing explanations with recommended actions. This makes each review more meaningful and efficient.
• Contextual policy guidance: The system provides employees with relevant policy information before they spend rather than flagging violations after the fact. This creates a proactive approach to compliance that reduces policy violations while improving the employee experience.

Transforming finance operations through intelligent automation

Brex’s AI updates have significantly impacted finance teams and employees. Now, Brex automates 60% of expenses, saving customers about 73,600 hours monthly with auto-generated receipts and expense documentation.

These benefits translate to significant financial impact, with customers collectively saving $56.5 million in salary dollars annually. For individual finance teams, auto-generated receipts and documentation save an average of $21,500 in salary costs and 204 hours per year.

Beyond time and cost savings, Brex's AI-driven approach creates a more effective compliance environment. While the industry standard for expense compliance hovers around 70%, Brex customers achieve a 94% compliance rate. This is possible through AI-driven anomaly detection that reviews 100% of transactions and identifies out-of-policy spend at the line-item level.

"Our goal is to empower employees to spend with confidence knowing that they are making appropriate business purchases without needing to be a finance expert," said Horn. "Let's be honest—almost no one reads their company's spend policy from cover to cover. AI can transform this experience by providing contextual guidance at the right moment, before the spending occurs. Instead of flagging violations after the fact, we can proactively guide employees with relevant policy information exactly when they need it."

This approach enables finance teams to focus on exceptions rather than reviewing everything, creating a trust-based system that scales. Foster said, "We would rather have people review by exception rather than review as the default."

Building an AI-first future for finance operations

Brex is innovating at the intersection of AI and finance, working closely with Anthropic and AWS to expand their Claude-powered capabilities in spend management. Their vision extends beyond simple automation to create a new paradigm where AI guides finance operations.

"Like many companies, we're enhancing our existing workflows by adding AI capabilities," said Horn. "But the future we're building toward is more transformative—where customers won't need to navigate our traditional interface at all. Instead, they'll simply have conversations with an AI that understands their needs and handles the financial operations automatically. For example, having a deep understanding of the company’s policies and SOPs so that all review and audit actions for clear-cut spend cases can be fully automated, escalating only the grey areas to humans for judgment calls."

This collaboration with Anthropic and AWS aligns with Brex's belief that as AI advances, the differentiating value in spend management will shift from software features to compliance expertise, partnerships, and intelligent automation—areas where Brex excels.

"As AI makes software development more accessible, technical features alone won't differentiate solutions," said Horn. "What will really matter is our expertise in compliance and our strategic collaborations—things AI can't replicate. Those are exactly the areas where Brex excels, and why we'll continue to lead in spend management."