🤖 Rox AI: A connector for Dialogflow ES/CX

This repository contains a dockerized distribution of Rox AI. Also, see Fonoster.

Youtube Demo

See a Car Rental demo using Rox AI

Available Versions

You can see all images available to pull from Docker Hub via the Tags page. Docker tag names that begin with a "change type" word such as task, bug, or feature are available for testing and may be removed at any time.

The version is the same of the Asterisk this is image is based on

Installation

You can clone this repository and manually build it.

cd fonoster/rox\:%%VERSION%%
docker build -t fonoster/rox:%%VERSION%% .

Otherwise, you can pull this image from the docker index.

docker pull fonoster/rox:%%VERSION%%

Usage Example

The following is a basic example of using this image.

docker run -it -p 3000:3000 fonoster/rox:latest

Deploying in development mode with Gitpod

One-click interactive deployment will familiarize you with the server in development mode.

Specs for Dialogflow backend

To allow for seamless integration between Dialogflow and Rox, we introduced the concept of Effects. Effects are actions sent from Dialogflow to Rox so you don't have to program the behavior every time. All you need to do is send the Effect's payload and Rox will react accordingly.

You can set multiple responses in Dialogflow. The Effects will run in sequence.

Effect ID	Description	Payload Example
say	The Effect will randomly pick a textual response and play it back to the user	{ "effect": "say", "parameters": { "responses": [ "Goodbye!", "Talk later", "Bye!", "Have a good one!" ] } }
hangup	The hangup Effect will close the call	{ "effect": "hangup" }
send_data	Use this Effect send arbitrary data to the client. Note that this only works with clients that subscribe for events	{ "effect": "send_data", "parameters": { "type": "map", "icon": "https://freeicons.net/icons/map.png", "link": "https://goo.gl/maps/YTum2VeZSQwNB4ik6" } }
transfer	Forward call to a different endpoint	{ "effect": "transfer", "parameters": { "destination": "17853178070", "record": true } }

Notes: The parameter type is set to map in the example, but you can send anything that makes sense to the client. If the parameter allRequiredParamsPresent is set to true, the fulfillmentText will take precedence over the custom effects.

Environment Variables

Environment variables are used in the entry point script to render configuration templates. You can specify the values of these variables during docker run, docker-compose up, or in Kubernetes manifests in the env array.

• DEFAULT_LANGUAGE_CODE - Sets the default language for the application. Defaults to en-US
• OTL_EXPORTER_PROMETHEUS_PORT - Sets Prometheus port. Defaults to 9090
• OTL_EXPORTER_PROMETHEUS_ENDPOINT - Sets Prometheus endpoint. Defaults to /metrics
• OTL_EXPORTER_JAEGER_URL - If set, it will send traces to Jaeger
• OTL_EXPORTER_GCP_ENABLED - If set, it will send traces to GCP
• OTL_EXPORTER_ZIPKIN_URL - If set, it will send traces to Zipkin
• EVENTS_SERVER_ENABLED - Activates the Events Server for socket connection. Defaults to false

Exposed Ports

• 3000 - Port to start a session request
• 3001 - Port to subscribe for send_data effects
• 9090 - Default Prometheus port

Volumes

• None

TODO

• Add authentication to secure the events port
• Include a --log-level flag (You can enable logs using the env LOGS_LEVEL)
• Include a --app-port so we can change the default voice application port
• Include a --events-port so we can change the default events port

Contributing

Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to us.

Authors

• Pedro Sanders

See also the list of contributors who participated in this project.

License

Copyright (C) 2023 by Fonoster Inc. MIT License (see LICENSE for details).