Skip to main content

JAICP updates in March 2025

New

  • LLM in telephony.
  • Get history in the LLM format.
  • Deploy bots to Mattermost.

Improved

  • $gpt.createChatCompletion method update.

Docs

  • Fields in $nlp.match response.

LLM in telephony

Beta

The llmRequest reply type has become available in JAICP. It lets you use LLMs in the phone channel with minimal pauses:

  • The bot receives text from the LLM by sentences and synthesizes speech by sentences as well. The two processes occur in parallel.
  • You can specify phrases that the bot will say to fill the pause at the beginning of text generation.

Get history in the LLM format

The new $jsapi.chatHistoryInLlmFormat lets you get the history of the dialog with the bot in the LLM format. For example, you can use this history in $gpt.createChatCompletion or in the llmRequest reply type.

Deploy bots to Mattermost

Mattermost is a messenger that is used as an internal chat in organizations and companies. It lets you create groups and channels, exchange files, and configure various integrations.

Now, you can deploy bots to this messenger via JAICP or Aimylogic. The Mattermost channel has the following features:

  • Send and receive audio, video, images, files.
  • Display buttons.
  • Launch text campaigns.
  • Transfer to agent.

$gpt.createChatCompletion method update

The $gpt.createChatCompletion method lets you access ChatGPT models from your bot script.

Now you can:

  • Select the model that will generate the response. GPT-4o mini is used by default.
  • Pass an object with any parameters accepted by OpenAI API. This provides flexible control over the model behavior.

Fields in $nlp.match response

We updated the article about $nlp.match and added a description of the new fields in the method response:

  • nBest
  • nluResults
  • intent
  • entities

With these fields, you can get more information about the classification results.