Inspiration
We have all experienced the frustration of buying something that does not work as expected, trying to understand a return policy, or spending too much time figuring out how to claim a warranty. Sometimes, people even miss return deadlines simply because they forget or do not know what to do next.
We wanted to explore how AI could make this process easier. Instead of just answering questions, what if an AI assistant could help people understand their options, prepare the necessary information, and keep track of the issue until it is resolved?
That is where ReturnRight AI started.
What it does
ReturnRight AI is a conversational assistant that helps people manage product returns, warranty questions, and purchase-related issues.
Users can describe a problem in their own words. The assistant helps them organize their purchase details, check the relevant return or warranty policy, understand their options, prepare a request, and keep track of important deadlines and follow-ups.
Our goal is to make a frustrating process simpler, clearer, and less time-consuming.
How we built it
We are designing ReturnRight AI around a conversational AI agent supported by tools that handle specific tasks. A Python backend coordinates the workflow, while an MCP server exposes the tools the agent needs to retrieve purchase information, check policy rules, and manage resolution cases.
We plan to use persistent storage so that users can return to an existing case without starting from scratch. A simple web interface will make it easy to follow the process and see what needs to happen next.
For the prototype, we can use sample purchase records and policy data to demonstrate the workflow without depending on private retailer systems.
Challenges we faced
One of the biggest challenges is making sure the assistant does more than generate convincing answers. Return policies can differ between products and retailers, so the system needs to rely on available policy information rather than make assumptions.
We also need to handle missing information, keep track of case status, protect user data, and ask for confirmation before any action that affects a real purchase.
What we learned
This project is helping us explore how conversational AI, tool calling, persistent state, and workflow automation can work together to solve an everyday problem.
We are also learning that building a useful AI product is not just about the model. Reliable data, clear user experiences, error handling, and user trust are equally important.
What's next
We want to test ReturnRight AI with real users, learn which parts of the return and warranty process cause the most frustration, and improve the product based on their feedback.
Our long-term goal is to build an assistant that helps people move from a confusing purchase problem to a clear next step, while keeping them informed and in control.
Log in or sign up for Devpost to join the conversation.