JAICP updates in October 2023
New
- ChatGPT integration.
- New project analytics:
- Revamped Project overview page.
- New
$analyticsservice methods. - Metrics configuration.
- New detailed reports.
- Speech synthesis markup in the Text block in J‑Graph.
- Default agent in Aimychat.
- Roles for working in Aimychat.
- Receiving data on bot phrase playback.
ChatGPT integration
JAICP introduces an integration with ChatGPT by OpenAI. You can use it, for example, to generate responses to unrecognized user phrases or to build creative solutions.
You can access the model directly from the bot script through the built-in $gpt service.
You do not need your own API key and OpenAI tokens to do this.
MLP tokens are already included in the JAICP plans.
For the model to generate responses to user phrases, call the $gpt.createChatCompletion method in the necessary state.
In the request, you can pass additional information about the dialog context and the user’s message history.
Then ChatGPT’s responses will be more accurate.
New project analytics
This month, we have totally revamped the JAICP toolkit for project analytics.
Project overview page
Now the Project overview page shows detailed statistics on the bot performance.
On the charts, you have access to summary data on the number of users, sessions, and messages and their changes by period. You can filter the values by the required periods and channels.
You can also analyze bot performance metrics:
- Percentage of automated dialogs.
- The evaluations that users give in the dialog.
- The results that the bot achieves during the dialog.
- The main topics on which the bot communicates with users.
New methods for collecting statistics
We have added new methods to the $analytics service so that you can get additional statistics on the project:
| Method | What it does |
|---|---|
setAutomationStatus | Records the dialog automation status. |
setNps | Records the dialog score. |
setScenarioAction | Records dialog actions. |
setSessionTopic | Records dialog topics. |
Metrics configuration
In the Analytics section, there is a new Metrics configuration page.
Here you can create new metrics: topics, actions, and session results, to collect data on these metrics in the script using the $analytics service methods.
setSessionResult method, you will find them on the Session results tab. The metric values will be the same as their keys.As long as the metrics do not have data in analytics, you can edit their keys or delete them. You can change a metric value at any time.
New reports in project analytics
In the Analytics section, there is a new Reports page.
Detailed reports collect data on user activity and script efficiency and statistics on custom metrics. For example, you can use it to see:
- How fast the audience of the bot is growing.
- How users rate bot performance.
- How often users receive an answer to their questions.
- How often the bot cannot solve a problem and transfers dialogs to an agent.
You can filter reports and charts for your business task and download them in the XLSX format.
Speech synthesis markup in the Text block in J‑Graph
In J‑Graph, it is now possible to use speech synthesis markup. Now in the Text block you can format bot phrases not only for text messages, but also for synthesis in the phone channel.
Default agent in Aimychat
Now you can assign a user as the default agent in an agent group in Aimychat. In the JAICP script, you can set the logic for transferring the dialog to this agent if they are offline.
Roles for working in Aimychat
In JAICP, there are new roles for working in Aimychat.
To add a new or existing Conversational Cloud user to Aimychat, you can assign them the AIMYCHAT_AGENT or AIMYCHAT_ADMIN role.
Receiving data on bot phrase playback
A new $dialer.getBotSpeech method allows you to receive data on bot phrase playback in the phone channel.
You can see at what point users stop listening to the phrases, analyze the results, and shorten some phrases. This data can be recorded in the script or as comments in analytics.