Share
In this Article
Share
At OnePay, we strongly believe that better money makes life better. We are so humbled by the opportunity to serve millions of customers, and wake up every day thinking about how to do more for our customers, faster. As I shared in The OnePay Platform Story, we have evolved from a single-product app into a comprehensive financial platform and growing. We are now reaching another inflection point. Generative AI has led to a fundamental shift in how we operate our business, build our products, and serve our customers .
This article is the first in a series where we will share our journey in deploying AI across the OnePay ecosystem. This post lays out the overall approach to AI at OnePay. We will share more details on each topic in future posts and as we continue on our journey.
We have focused our AI efforts in three distinct but interconnected areas: Operations, Productivity, and Product.
1. Operations: Maximum Efficiency in Customer Service
Our objective is to service customer contacts as quickly as possible, and with the highest degree of quality, while maintaining extreme operational efficiency. To achieve this, we have integrated AI deeply into the full lifecycle of the customer support journey. This isn’t about AI bots, it’s about the end-to-end lifecycle of the customer experience.
We have developed and deployed five specialized AI agents across three phases of the customer support lifecycle:
Before Contact: Chat and Phone Agents serve as the first line of contact. Their goal is to provide a high-quality, personalized service in a natural conversation. They handle routine inquiries instantly, allowing our customers to get back to their day.
During Contact: We recognize that some customers might require more support than AI is able to provide, and we route them to a representative quickly. We’ve deployed the Co-Pilot Agent, an AI agent that works alongside each representative. It follows the representative conversation with a customer, retrieves information, suggests solutions and helps resolve cases faster and with higher quality.
Post Contact: The work doesn't stop when the conversation ends. We developed a Quality Assurance (QA) Agent that reviews every contact after it concludes, ensuring we meet our high standards and suggests specific areas for improvement. Simultaneously, our Analytical Agent reviews and summarizes contacts to uncover invaluable insights and trends, directly informing our product decisions.
2. Productivity: Moving at the Speed of AI
The second area of focus is our own productivity. The objective is simple: enable every function at OnePay to move even faster.
We have deployed a suite of general-purpose AI agents that integrate directly within the tools we use across the company. These include agents that attend meetings to capture notes and follow-ups, operate within Slack and email, and assist with documents and spreadsheets. Employees also have access to an agentic research assistant, and we’re rolling out a workspace that enables each employee to build and manage their own agentic workflows.
We also deployed a (growing) set of specialized AI agents, tools and Model Context Protocol (MCP) servers to supercharge specific business verticals.
In Engineering, we found that while AI is exceptional at writing code, that capability barely scratches the surface of what’s possible. We realized that one of the largest datasets we have is our codebase itself, featuring millions of lines of code.
We developed a specialized Developer Agent that understands this codebase, and can engage in real-time natural conversations about it with anyone at the company. The agent can propose code changes, visualize those changes and submit pull requests for review.
We built an Incident Agent that helps us manage incidents and post mortems, a Pipeline Agent that connects to our build and release tools to manage and debug release pipelines and a Code Quality Agent that performs code reviews aligned with our coding standards. Several other agents are currently in development.
As a result of this evolution, alongside other platform enhancements, the number of changes we deploy to production has increased by more than 5x over the past 12 months.
In other functions, specialized AI agents are helping us source and recruit candidates, perform customer research, manage concurrent procurement processes, and assist in various other business functions.
By equipping our teams with these capabilities, we free up time for higher-value strategic work, allowing us to focus more directly on solving our customers’ most important needs.
3. Product: Simple, Helpful, and Delightful
The third and perhaps most visible area of focus is the use of AI in our products. Our objective is to create simple, helpful, and delightful experiences everywhere OnePay shows up for our customers.
We started by identifying where AI could simplify existing interactions for the millions of customers who use our app every day. But we decided to take it a step further. We developed the OnePay Money Companion (OPMC), a personal AI that lives directly in our app. Users can chat with their companion in a simple, natural way. Each conversation is personalized and contextual. OPMC is currently rolling out.
As our customers delegate more of their daily tasks to AI agents, we believe our products must also cater to those agents working on the customers’ behalf. We recently announced that we have joined Google’s Agent Payments Protocol (AP2). This is a prime example of how we are shaping the future of agentic payments, ensuring that when AI agents transact, they do so securely and transparently. You can read more about our work with Google and AP2 here.
AI as a Core Platform Capability
None of this would be possible without the architectural foundation we laid out in our Platform Story. We realized early on that to move fast, we couldn't build bespoke solutions for every use case, and AI was no different.
We have invested in building a reusable AI layer within our platform. This layer enables our various use cases across products and in operations by providing shared services for managing knowledge bases, hosting MCP servers, and a pluggable backend that allows us to rapidly switch between different large language models, LLM versions and providers, ensuring we are always using the best model for the job without rewriting code.
By treating AI as a platform capability, we ensure that building AI features can be done quickly, leveraging a robust, secure, and scalable infrastructure.
Looking Ahead
This is just the beginning. The roadmap ahead is filled with opportunities to help more people, and the pace of innovation is accelerating. Over the coming weeks, other members of the OnePay team will publish articles diving deeper into these specific areas, from how we built our support agents to the technical details of our internal tools and the AI features we roll out to customers.
Stay tuned.