From legacy to leading edge: Expensify’s AI-powered expense management framework

I've always believed that the opportunity before us is enormous. Today, I want to share our framework for how we’re taking legacy business processes and transforming them with AI.

At its core, our approach is simple: build a chat-centric experience that seamlessly marries our deep, proprietary data with the superintelligence of AI.

Let me walk you through our three-step process—Deep AI, Surface AI, and Elevated AI—and explain how this framework sets the stage for a new era in financial technology.

The chat-centric revolution

AI isn’t just about algorithms and machine learning, but communication. When you think about it, the language of AI is English. Chat is how we interact daily—whether texting with friends or managing work emails. That’s why we designed Expensify Chat and our Concierge system.

Unlike generic AI models that know everything about the public world, our Concierge is built to understand your private, contextual data. It’s not about replacing existing tools like Slack—it’s about embedding intelligence directly into your workflow.

With a chat-centric interface, we’re not just answering questions; we’re initiating conversations that matter, right where you need them most.

Chatbots that actually work

"Concierge has helped reduce banking problems by 75%."

Read more about David's vision for AI-driven expense management here.

Step 1: Deep AI – building on legacy data

The journey begins with Deep AI. Think of it as the foundational layer where we train our models on every bit of historical, human-generated data we’ve accumulated over 15 years. This data isn’t just numbers—it’s a record of countless receipts, expense reports, and user interactions.

By harnessing this rich dataset, we teach our AI to master even the most minimal judgment tasks, like accurately scanning receipts or processing expense entries. Our AI receipt scanning technology ensures that expenses are captured with pinpoint accuracy, reducing errors and minimizing the need for manual corrections.

This “low-hanging fruit” creates huge cost savings and builds a robust base for all the AI enhancements to come.

Step 2: Surface AI – enhancing the user experience

Once our AI understands the basics, we move to Surface AI. Here, our trained models evolve from simply processing tasks to interacting with you in a meaningful, conversational way.

Imagine a scenario where you submit an expense that might be ambiguous—a $7 entry labeled “Subway.” Instead of guessing, our system engages you in a natural chat, asking clarifying questions to determine whether it’s for a meal or transportation. This layer leverages our extensive domain knowledge to offer bespoke responses, making every interaction personalized and efficient.

With AI expense tracking, our system continuously monitors spending patterns, categorizing and organizing expenses in realtime, ensuring accuracy while streamlining financial workflows. Surface AI is all about turning data into dialogue, improving accuracy, and reducing the need for manual oversight.

Step 3: Elevated AI – Superhuman insight and proactive management

The final step is Elevated AI—where we unlock the true potential of our system. This isn’t about doing what humans can do faster; it’s about doing what humans can’t do at all. Elevated AI takes our chat-driven foundation and scales it to perform advanced analyses in realtime.

Whether detecting unusual spending patterns before they become a problem or offering proactive advice like a virtual CFO would, this level of AI operates at a superhuman scale. It monitors your AI expense tracking data continuously, flags anomalies, and even integrates with external systems to pull in additional context.

In short, Elevated AI isn’t just reactive—it’s anticipatory, giving you the tools to manage your business smarter and faster than ever before.

The future is now

We’re not just making incremental changes here. We’re rethinking how legacy businesses operate by embedding AI deep into every process. With our chat-centric design, every employee—whether in finance, sales, or the C-suite—can interact with an AI that understands the data and the context unique to your business.

The transformation we’re pioneering isn’t about replacing people; it’s about empowering them to focus on higher-value tasks while our AI handles the repetitive work.

This framework of Deep AI, Surface AI, and Elevated AI is more than a roadmap—it’s our commitment to driving a total fintech AI to primacy. We view the tidal wave of AI not as a threat but as an opportunity to reshape how businesses operate, creating efficiencies and unlocking new levels of insight along the way.

Welcome to the future of expense management and beyond. It’s an exciting ride, and I’m thrilled to have you join us on this journey.