Generative AI : Proposition de Enriched response suggestions with InboxCare

InboxCare combines deterministic responses with adaptive capabilities and tonal creativity of the latest LLM technologies.

Why this approach

Deterministic responses ensure reliability and are essential for requests where hallucinations are not allowed. LLMs, on the other hand, provide empathy and tone adaptation and are ideal for interactions requiring a human touch. Combined together, they ensure replies that are both accurate and personalized.

After analyzing and categorizing the message, InboxCare generates a response based on the identified categories. For some categories, a predefined text is used, while for others, an LLM is called to enrich the response. InboxCare also automatically adds a predefined header (e.g. "Hello") and signature (e.g. "Best regards, Your Customer Service"), ensuring consistent and personalized responses to every interaction.

How it works

Set a response for each message category

Each message category has a corresponding template. A template with predefined text, or a template that includes a prompt that will be sent to an LLM.

You can also define templates that you want to appear on each response (such as greetings).

💡 We use Mixtral 8x7B as the LLM for our response proposals. Mistral is a French company based in Paris and regulated by the European AI Act.

InboxCare links each response with the categories found

InboxCare analysisanalyzes the message, detects the categories, then builds the response based on the categories found:

If a category has a predefined text connected to it, InboxCare will add the text as is to the final response.
If a category is associated with an LLM prompt, InboxCare will send the prompt to the LLM and add the received response to the final reply.

InboxCare sends the generated response via the tool of your choice

Once the response proposal has been built, InboxCare delivers it via the channel of your choice:

In a messaging tool, like Gmail or ExchangeGmail or Exchange messaging tools
In a ticketing tool like JIRAJIRA or other ticketing tools
In a CRM tool, like Salesforce or HubSpotSalesforce or HubSpot CRM tools
In a JSON format so you can integrate it into your information systemJSON format so you can integrate it into your information system

Ready to go?

Try out our powerful, explainable and adaptable AI.

Generative AI : Proposition de Enriched response suggestions with InboxCare

Why this approach

How it works

Set a response for each message category

InboxCare links each response with the categories found

InboxCare sends the generated response via the tool of your choice

Ready to go?

Subscribe to our newsletter

PRODUCT

SOLUTIONS

TECHNOLOGY

COMPANY

RESOURCES

© 2024 Golem.ai

Follow us