Coach Call AI

• 

  Coach Call AI

Inspiration

As someone with ADHD, I always struggled to wake up on time and feeling motivation in the morning to get to work. I always wished there was a friend who would call me every morning after my alarm rang to give me a pep talk to get out of bed. This is how Coach Call AI started. As it progressed, feedback from users demanded that I also extend it to a best friend that checks in, hence the 3 different modes we have today.

What it does

Coach Call AI is the first AI companion you can talk to via phone calls and WhatsApp, keeping you accountable with regular check-ins and meaningful conversations to help you stay on track with your goals and life. Whether you need accountability coaching, mindfulness guidance, or a completely custom companion, your companion adapts to your specific needs through three specialized modes and through long term memory. It proactively reaches out via WhatsApp throughout the day and makes actual phone calls when you need that extra push or support.

How I built it

I built Coach Call AI using React and Vite, integrated with Supabase for real-time data management and user authentication. The core innovation lies in the integration with VAPI for natural voice conversations and WhatsApp's Business API for seamless messaging. I developed three distinct AI modes (Accountability, Mindfulness, and Custom) that users can switch between, each with specialized prompts and personalities. The system remembers every conversation and learns from user interactions to provide increasingly personalized support. This works through an AI that flags important messages, another that extracts and logs achievements to a database, and several background tasks. These tasks update the user’s profile with a summary of key messages, send check-in messages via WhatsApp, and trigger scheduled phone calls when it's time to call a user.

Challenges I ran into

The biggest challenge was creating AI that felt genuinely human in both text and voice interactions while maintaining consistency across different communication channels. I also had to account for different time zones when managing scheduled calls and WhatsApp check-in messages, making sure everything happens at the right time for each user, and ensuring the AI remembers context across weeks or months of conversations were significant technical hurdles. I also had to tackle the challenge of making the AI smart enough to learn about each user and potentially schedule calls automatically through WhatsApp (though this feature is still in progress), all while keeping the service affordable. It started out as a simple accountability coach, but expanding it to support three distinct modes while keeping everything flexible and seamless on the backend was a challenge, especially when handling users switching between modes, each with their own scheduled calls. The system had to ensure that only the calls relevant to the user’s active mode would go through.

Accomplishments that I'm proud of

I've successfully created the first AI companion that seamlessly bridges voice calls and messaging, making accountability feel like having a real friend rather than using an app. Even before I officially launched or promoted the app, some users started finding and using it organically through search engines. The three-mode system allows Coach Call AI to support a wide range of users, from busy professionals who need accountability, to those looking for daily mindfulness, or simply a friend to talk to.

What I've learned

Beyond the technical lessons I gained while building this, like database security, edge functions, and techniques for managing both short and long-term memory, I’ve also learned something more human: people truly need someone to check in with them, and many deeply crave connection. As shown by the success of companies like Replika and Character.ai.

What's next for Coach Call AI

At the end of the day what matters most is user feedback. I'm staying flexible and open, ready to steer the project in whatever direction best serves my users. The roadmap will evolve accordingly since user needs are the top priority, and I’m always actively collecting feedback.

Right now I’m expanding the AI’s emotional intelligence to better recognize when users need different types of support. I’m also building deeper integrations with health and productivity apps to offer more comprehensive life coaching. And my current priority is creating an intelligent system that lets users schedule calls using natural language directly through WhatsApp, without needing to type commands like “/call” or visit the website.

My goal is to build the AI companion that doesn't just help you achieve goals, but helps you become the person you want to be, available 24/7 through the platforms you already use daily.

Built With

• bolt.new
• chatgpt
• react
• stripe
• supabase
• tailwind
• typescript
• vapi
• vite