Arc AI

Inspiration

Some of us are founders and have experienced the pain of customer discovery from both sides. As founders, we struggled to reach users and gather meaningful insights; as customers, we often endured long, repetitive surveys that felt like a waste of time. Experiencing this tension made us realize there’s a gap in how businesses communicate with people, and that interactions need to be smoother, faster, and more human.

What it does

The product automates outbound phone calls using AI voice agents to conduct customer surveys and discovery calls at scale, then captures and analyzes the conversation data to provide actionable insights.

How we built it

We started with a discussion to define the problem and identify which product would fill the biggest gap. The base of our prototype was built on Lovable, but we used Vappi to handle natural, human-like voice interactions and Twilio for outbound phone calls. To make the system flexible, we created multiple AI personas for different use cases, including a customer interviewer persona, using targeted prompt engineering. Each conversation was captured as a transcript, which we processed through an LLM to extract insights, sentiment, key themes, and actionable items

Challenges we ran into

• Workspace Access & Sharing: Managing permissions and sharing code was tricky, and we frequently ran out of credits in a single workspace, which forced us to create new workspaces.
• Real-Time Code Review: The platform isn’t developer-focused, so we couldn’t see the terminal or internal processes.
• Limits on Complexity & Feature Integration: Combining multiple components pushed the platform’s limits for how much complexity and integration it could handle.

Accomplishments that we're proud of

We’re proud of building a solution with a highly impactful main use case that addresses real customer and business needs. Our team successfully collaborated across diverse backgrounds, bringing together members from four different programs to create a truly multidisciplinary approach. We also developed real-time voice interactions and proactive outbound calling, demonstrating a functional, scalable, and AI-powered workflow.

What we learned

As we explored the space, we found the same pattern everywhere: everyone is frustrated. Businesses struggle to gather reliable insights from customers, and individuals often feel like they’re giving their time without receiving anything meaningful in return. Today, most experiences still feel impersonal, repetitive, and transactional. We also realized this gap isn’t limited to customer discovery, people want affordable, accessible support for things like interview prep, language learning, educational coaching, or even therapy-style emotional check-ins. Across both business and consumer use cases, our biggest learning was that the real barrier isn’t automation, it’s motivation. People only engage when the interaction feels genuinely useful and when they clearly understand the value or support they’re getting.

What's next for Arc AI

If we had more time we'd go-to-market with our MVP for outbound therapy, then build up B2C capabilities before taking our first steps into B2B markets. We'd then expand to include both inbound and outbound solutions, tailoring the inbound solution based on what we learn from our therapy MVP. We think about scaling this by gathering customer feedback on how the platform works and continuously iterating. We'd start with therapy to validate our conversational AI, strengthen our B2C offering, then leverage that proven technology to scale into B2B customer discovery and support use cases.

Built With

• grok
• twilio
• vapi