Inspiration
Advan was inspired by a frustrating customer support experience. I once spent more than 20 minutes on hold, was transferred between multiple departments for a simple refund question, had the call drop, and then had to call back and start the process again. That experience made me think about how much worse this problem becomes for growing businesses when customer demand increases faster than their support teams.
What it does
Small companies often have lean teams supporting a growing customer base. Customers expect quick, accurate answers, but support teams can become overwhelmed. Existing AI support platforms often focus on large enterprises, leaving smaller companies with fewer affordable, flexible, and transparent options. We built Advan, an AI-powered customer support platform designed for growing teams. Advan combines AI automation with Human + AI Copilot workflows so teams can automate routine support while keeping humans involved when situations require judgment. Our Tap Box adds a transparency layer to AI responses by showing confidence scores, cited sources, and policy validation. When the AI is uncertain, the workflow can route the conversation to a human instead of confidently inventing an answer, which is apparently still something AI systems occasionally enjoy doing.
How we built it
We started by researching the customer support challenges faced by growing businesses and conducting customer discovery interviews. We then built a working MVP focused on the core support workflow: Customer question → retrieve relevant context → generate response → verify response → confidence check → AI response or human escalation We developed the product around rapid experimentation and feedback, refining the workflow, user experience, and transparency features as we learned more about the problem.
Challenges we ran into
Building as an early-stage startup meant balancing product development, customer discovery, marketing, and sales with limited resources. One of our biggest challenges has been finding the right early users and validating product-market fit. Rather than trying to build everything at once, we focused on a narrow MVP, gathered feedback, and prioritized the features that could create the most value for early customers.
What we learned
One of our biggest lessons was that AI support is not simply about generating better answers. Trust is equally important. Businesses need to understand why an AI system produced an answer and have a reliable way to intervene when it is uncertain. We also learned that small teams need solutions that are simple to adopt and provide clear value without the complexity of enterprise software.
What's next for Advan
We are now recruiting more users and design partners to test Advan in real customer support workflows. Our goal is to continue learning from these users, improve the product, validate product-market fit, and make trustworthy AI support accessible to growing businesses.
Log in or sign up for Devpost to join the conversation.