LLM Apps

What is an LLM app?

An LLM app is an intelligent workflow that orchestrates different processing steps, combining the power of language models (LLMs) with the specialized services of the Wikit Semantics platform. This orchestration makes it possible to create rich and contextualized user experiences.

How it works

Each LLM app follows a sequence of steps optimized for its use case. For example, in the case of a conversational question-and-answer application, the typical workflow includes:

• Semantic search for relevant fragments in document sources
• Identification and traceability of sources used
• Invocation of the LLM to generate a precise and contextualized response
• Management of informal conversations (small talk)
• Validation and formatting of the final response

Types of LLM apps available

Wikit Semantics offers several pre-configured LLM app templates to meet different needs:

Personal assistant

• Access to up-to-date information from the web
• Transparent online source citation
• Upload PDF, DOCX, XLSX, … documents in Semantics Chat
• Personalized starters to facilitate interactions

Question/Answer Chat

• Management of multi-turn contextualized conversations
• Intelligent leveraging of data sources
• Maintenance of conversational context
• Ideal for virtual assistants and customer support

Simple Question/Answer

• Direct answers based on document sources
• Single-question mode (without conversational history)
• Perfect for FAQs and ad-hoc information retrieval

Each LLM app type is optimized for its specific use case, ensuring optimal performance and a user experience tailored to needs.