character3d is a framework that lets you have a 3D talking head chat as the interface for your AI endpoint.

as long as your endpoint returns the response below, the 3d talking head will be able to process it

The user sends data posted a form variable called query (for the hackathon, please do NOT use json... just imagine if your endpoint accepted a input field name="query"), equivalent to this web form post:

<form method="POST">
<input name="query" value="[user chat]" />
</form>

And the app expects a response in this format:

json version of it (replace values with your own, not the commented dummy text)

{
    "query": " // nice to have (the original query the user sent)  ",
    "txtresponse": "// required ",
    "urlpolly: " // url to a MP3 file - optional, by default AWS Polly will be used to text to speech"
}

sequenceDiagram
User ->> App: things to say to AI
App ->> Server: post a query variable to endpoint
Server ->> App: returns GeneralizedEndpointResponse json 
App ->> User: 3d talking head

Here is the order of prefrence of content for the talking head:

1. urlpolly - if you have a mp3, the talking head will animated from that
2. if not, then we will generate it from txtresponse