MoM Extraction (AI Integration)

[ndhanvina]

• Design MoMExtractor class (OOP)
• Define class structure and methods for extraction.
• Define prompt template for MoM extraction
• Write and test prompt for extracting MoM sections.
• Integrate LangChain with Ollama
• Set up LangChain to use Ollama as the LLM backend.
• Implement extraction logic
• Code the method to send transcript and receive structured MoM.
• Write unit tests for MoMExtractor
• Test extraction with sample transcripts.

[ndhanvina]

The mom_extractor.py file (inside the extractor/ folder) is responsible for the core AI logic:
It takes a meeting transcript as input,
Uses LangChain and Ollama to interact with an LLM,
Extracts and returns the structured Minutes of Meeting (MoM) sections.
What Should Be in mom_extractor.py?

1. MoMExtractor Class
   Initialization: Set up the LLM (Ollama via LangChain), and load or define the prompt template.
   Extraction Method: A method (e.g., extract_mom) that:
   Accepts a transcript (string),
   Sends it to the LLM with the prompt,
   Parses and returns the structured MoM (as a dict or custom object).
2. Prompt Template
   The prompt that instructs the LLM to extract the required MoM sections.
3. (Optional) Helper Methods
   For parsing, error handling, or post-processing the LLM output.

@anushacybrisk

[ndhanvina]

@anushacybrisk prompt template

[ndhanvina]

@ndhanvina ollama llama3 model class