Integrating into GenerativeAgent requires you to hook GenerativeAgent into your voice or chat system. This enabled GenerativeAgent to talk to your users.

An end-to-end integration of GenerativeAgent can be represented by these key components:

  • Connecting your data sources and APIs.
  • Listening and Handling GenerativeAgents. Create a single SSE stream where events from all conversations are sent.
    • Events are how GenerativeAgent’s response is sent, as well as other key status information.
  • Send your audio or text conversation data to ASAPP and have GenerativeAgent analyze the conversation.

Passing the conversation data, analyzing with GenerativeAgent, then handling events functions as a loop until the conversation completes or the conversation needs to be escalated to an agent.

The below diagram shows how these components work together, and the general order in which they execute during a conversation:

  1. Create an SSE stream and handle the GenerativeAgents events sent via the stream. GenerativeAgent’s reply is sent via this event stream. You need to provide the bot’s response back to your user.
  2. Send your audio to AutoTranscribe. Use one of our Connectors to streamline this integration. Otherwise, you can use our websocket integration to pass raw audio.
  3. Pass the conversation transcript into ASAPP. This is the transcript you receive from AutoTranscribe or the direct text of a conversation if a chat channel like SMS.
  4. Engage GenerativeAgent on the conversation with the /analyze call. GenerativeAgent will look into ASAPP’s conversation data to account for the entire conversation context.

You need to configure GenerativeAgent in order for it to connect to your Knowledge Base and APIs.

​
Connectors

With many IVR or contact center setups, you will not have easy means of direct audio capture. We support several connectors to make integrations easier:

MRCP

We have a plugin for Uni for most On-Prem IVR contact center solutions.

Coming soon:

  • Amazon: For contact centers build on Amazon Connect
  • Genesys: For contact centers powered by Genesis Cloud.

If there is a platform we do not support that you would like added, please reach out to your ASAPP team.

If there is a platform we do not support that you would like added, please reach out to your ASAPP team.

​
Direct API

We also support a more direct integration into GenerativeAgent:

Audio via AutoTranscribe

Use our AutoTranscribe to transcribe your user’s audio. You would be responsible for converting GenerativeAgents text back into audio.

Text Only

Send the text of a conversation directly to GenerativeAgent. This is for Chat based systems, or if you are handling all your own transcription and Text-to-Speech.

​
Examples

We have various examples of interactions between a user and GenerativeAgent to show what API calls to make, and what events you would receive.

Each integration method has it’s own uniqueness, but these examples should still generally apply.

​
Next Steps

With a functioning GenerativeAgent integration, you are ready to call GenerativeAgent and receive analyzed replies.

You may find one of the following sections helpful in advancing your integration:

Handling GenerativeAgent Events

Text-only GenerativeAgent

Go Live

Was this page helpful?

Safety and TroubleshootingUniMRCP Plugin for ASAPP