Inspiration

The project is inspired by monthly calls and WhatsApp messages I get from my car cleaner and newspaper vendor to send them their payments.

What it does

It allows anyone who needs to make money calls on a monthly basis to more than 10 people a breeze. They can press one button and it makes the call.

The app allows them to add their customers and the amount to be collected each month. They can press a button to generate a script that they like (uses Claude) in Hindi, English or Kannada (any language can be added very easily with AI), and they select voices available to them (eleven labs).

It doesn't overdo things. It doesn't send payment link, so payment flow is the same as before - directly to the service providers. Service Provider manually indicate customers who have paid, and can make the call again for the remaining customers.

How we built it

The project has been on my mind for some time but I built it after seeing the Hackthon post. It's built using Cursor, Claude, Gemini to code. Stack includes Railway, Firebase, AWS, ElevenLabs and Knowlarity. The project is 100% "vibe-coded". I'm a PM turned full-time vibe-coder.

Challenges we ran into

  • Nothing significant to be honest (AI knows everything, especially if requirements are super clear)
  • Documentation: I expect documentations to be correct and available in .md.
  • Context window issue. Also, ran out of monthly credit on one LLM so had to switch to other.

Accomplishments that we're proud of

  • It works from registration to making the phone call (there's some issue with receiving Webhooks from Knowlarity so we can't update Campaign status accurately) - that's nice to see.
  • AI brings down the cost of building software significantly so on can now think of building for otherwise underserved use cases.

What we learned

  • Learnt about Eleven Labs and Knowlarity since it was my first time using them

What's next for Capital Call

  • Get its first 20 customers

Built With

Share this project:

Updates