The next generation of intelligence built on NLU/NLP AI

Azure AI/ChatGPT Conversational Service


Decrease training cost, smarter conversational service, higher accuracy

Intumit SmartLLM product integrates LLMs to create cutting-edge artificial intelligence products based on NLU/NLP. Utilizing Azure AI/ChatGPT models, our customers can leverage the latest NLP technology to automatically generate training data, detect intents, and generate initial commands. This streamlines the conversation building process, reducing time and manpower requirements. With less training data but higher accuracy, we aim to deliver superior conversational services.

To enhance both customer and employee experiences, Intumit focuses on providing improved predictiveness, applicability, and versatility in the development and tuning of its SmartLLM product. We aim to deliver advanced generative AI services that support multiple languages, enabling multinational brands/companies to create chatbot solutions at lower costs.

SmartLLM with Azure Open AI / ChatGPT

In order to improve customer and employee experiences, Intumit prioritizes enhancing the predictiveness, applicability, and versatility of its SmartLLM product during development and tuning. Our goal is to provide advanced generative AI services that can accommodate multiple languages, thereby enabling multinational brands/companies to develop chatbot solutions at reduced costs.

Main Function


Establishing a structured FAQ knowledge base

Transforming static frequently asked questions (FAQs) into a specialized structure based on key domain terms. Connecting them with context-specific questions, while also including their alternative questions, synonyms, and enabling machine learning features.

  • Accelerating the creation of FAQ knowledge base questions: Upload document or article content to automatically generate training questions and sample sentences.
  • Accelerating the creation of FAQ knowledge base answers: Utilize the powerful summarization, induction, and expansion capabilities of LLM to alleviate the burden of editing answers.
  • Accelerating the creation of FAQ knowledge base vocabulary: Automatically generate synonyms.

Combining answers for multiple intents

The user poses multiple questions in a single inquiry, and the chatbot can retrieve multiple matching answers from the FAQ knowledge base. It then utilizes LLM to blend and generate a response that meets the customer’s expectations, effortlessly handling various challenging and complex queries.


Semantic Search

Using Azure OpenAI integrated large language model (LLM) to generate answers by processing uploaded documents and user queries. It helps users answer questions from unstructured text documents without the need to create individual FAQs and train them.


Azure OpenAI directly provides reference answers

Using Azure OpenAI integrated large language model (LLM), users can generate reference answer content by providing only the question, with settings for tone, role, and direction of the response.




Parse voice messages

Upload STT, LINE CALL voice files, or other voice system files to the platform for voice content parsing, and generate key summaries or Q&A based on the parsed content.


Optimize answer content

Optimize answer content for tone, different language variations, response methods, and more.


Build chatbots quickly with fewer training resources

Integrate traditional flyers, reports, registrations, and email notifications into a single channel

Execute extensive and complex tasks through AI-powered conversations

Handle multiple languages without the need for translation personnel

Learn complex structural logic like a human

Design marketing and shopping guide processes

Shorten service development time

Reduce expenditure costs

Obtain more precise answers



The text files that can be uploaded include Excel files for building Q&A documents. For automatic generation of QA, the upload formats include pdf, docx, pptx. The voice files include STT, LINE CALL voice files, and other voice system files.

In the prompt, you can set the content and scope of the response. You can instruct not to mention competitor-related or specific brand-related information. Additionally, when standard or similar questions cannot be found, you can set the requirement for GPT to reply based on other conversation guidelines.

SmartLLM can set the tone of the answers and provide multilingual answers. It can also automatically generate answers based on questions, enabling quick delivery of the best answer content.

Implementing SmartLLM is mainly divided into four stages. The first stage involves brainstorming initial prompt based on the industry type. The second stage involves explaining the ChatGPT testing method. The third stage involves testing by clients and feedback on the testing form. The fourth stage involves adjusting prompts and knowledge Q&A based on the testing feedback.

For personalized service explanation, please fill out the form below to Request a Demo and contact our team.

Building the next generation of intelligence based on NLU/NLP AI

Azure AI/ChatGPT Conversational Service