Gemini API Developer Competition

Video Demo for the Gemini API Developer Competition 2024
#buildwithgemini #gemini

How it works
It's a backend that receives text from speech of phone call, transforms into actions and execute those actions. The flow is:
User call the number and give it a action ("Call the local pharmacy and check if they have headache medicine")
The system, using Gemini API, asks all the needed informations for the action (like names, phone numbers, address...)
When the system has all the info, the call ends
Then the system review the call (using Gemini API) and transform into actions (and take notes from important info)
After that, the system execute the action (make a outbound call) and talk to target person of action
When the system get the information needed to consider the action completed, the outbound call ends
Finally, the system call the user back with the summary from the action

The gemini API is used to process all the text/chat:
The text from the phone call
The reasoning for reviewing and transforming the chat into actions
The reasoning for reviewing and transforming the chat into notes
The reasoning for reviewing and transforming the chat into summary for the return call