Odichat

No-code web platform to automate customer service tasks in WhatsApp. Create and train customer support AI agents that integrate into WhatsApp Business Accounts to automate answering FAQs.

Comment

Inspiration

As a customer, I get upset and frustrated when a business is unresponsive.

But on the other side, a business might receive dozens, if not hundreds, of daily customer questions/inquiries at any time.

It's inevitable: leads slip...

On the other hand, as a customer, I hate IVR systems, but most importantly, I'd rather text than call for support (probably a Gen Z thing?).

So I built Odichat.

What it does

A no-code web platform to train and deploy customer support AI assistants to WhatsApp Business

How we built it

Odichat was built with Ruby On Rails 8, StimulusJS, TailwindCSS, SQLite3, Digital Ocean (for storage), and deployed on Hetzner using Docker and Kamal 2.

Challenges we ran into

There are three main challenges I've faced building this app:

  1. Becoming Meta's official Tech Provider
  2. Learning to navigate the Meta for Developers platform and the Meta Cloud API.
  3. Learning about LLM APIs

Accomplishments that we're proud of

Becoming Meta's official Tech Provider and learning how to navigate their docs is what I'm most proud of in this project.

What we learned

If I were to highlight one thing is that building a SaaS platform is HARD. I've learned a lot of things along the way (and keep learning every day).

  1. Time management
  2. Database design & backend
  3. UXUI design & front-end
  4. Security
  5. LLMs and AI
  6. CI/CD
  7. Testing
  8. So many other things I can't even list here...

What's next for Odichat

  1. Launch & gather user feedback & fix bugs (priority)
  2. Extend AI Agent functionality with tools (e.g.: Calendly integration, Zendesk, etc.)
  3. Integrate analytics, so users get insights and visibility on how their customers are using their AI Agents (spot patterns like FAQs and customer satisfaction)
Share this project:

Updates