A platform which provides 24/7 availability customer supports, act as a spam/scam blocker for businesses or individual, acts as your AI receptionist or sales representative / call centre.
A platform which enables businesses integrate their business and provides a way to automate the process of either:
- Booking an appointment
- scheduling a meeting
- Placing an order ( a node base UI would be use to plan the automation. )~~
- Answer questions and answers based on the resources uploaded from the business or used the data provided by the businesses
- Provides a chatbot embaddable widget
- Set available time ( necessary especially if businesses need to prevent user from scheduling a meet on a day they aren’t available )
It also Provides a sell phone number users could call to perform some of the functions above.
Businesses could set some configurations as well which tells the AI if a caller or user could attempt an action. (still giving this a thought)
- Web ( Main App )
- React + vite
- Chartjs ( document / graph out the intent of each calls or message made )
- Chatbot widget
- React + vite ( would be accessible via an iframe using app deployed url )
- Nodejs / Expressjs ( core backend api )
- Postgresql ( supabase )
- Inngest ( background jobs )
- Gemini AI API
- I dont think this are needed but would leave them here.
- TTS ( google )
- STT ( google / openai whisper )
- Integrations
- meet
- calendar
- Twillio
- sending sms
On signing up users are redirected to onboarding screen (if they dont have any business created) which asks for the following:
- name
- type
- phone number
- email (optional)
- opening days
- opening hours
- closing hours
- country
- Metrics
- Total calls
- Total messages
- Total appointments
- Sentiment analysis ( Graph ) This could be filtered based on either (conversation between AI and user (in-call) ) or chat messages (chatbot)
User could choose to either use the chatbot or call the phone number provided.
- Chatbot widget
User could upload resources which would be used by the AI to answer questions asked by extrernal users. each bucket provides a way to group resources together. (website, docs, pdfs, etc)
- name
- description
- Data
- web url
- custom data (txt)
This includes both chat messages and call messages (messages extracted from the call btw the AI and the user)
- Sentiment Analysis Graph ( Still giving this a thought )
- provide sentiment analysis of the conversation of the customers response messages.
- Type
- chat (chatbot)
- call (phone call)
Contains all the appointments made by the customers.
- date
- time
- user
- status (pending, completed, canceled)
Add third party integrations.
-
Whatsapp (still giving this a thought) provides a way to send messages to the user via whatsapp since sms could be expensive.
For converting site content into valid markdown, we could use html-to-markdown npm package. Or perhaps using a small cloudflare LLM worker to convert the content into markdown cloudflare LLM