Skip to main content
Back to Library
Prompt Engineering Guide

Mastering Meeting notes extraction
on Qwen 2.5 72B

Stop guessing. See how professional prompt engineering transforms Qwen 2.5 72B's output for specific technical tasks.

The "Vibe" Prompt

"Extract key points, action items, and decisions from these meeting notes: [MEETING_NOTES_TEXT]"
Low specificity, inconsistent output

Optimized Version

STABLE
You are a highly efficient and accurate meeting notes summarizer. Your task is to extract specific, structured information from the provided meeting transcript. Follow these steps meticulously: 1. **Identify Key Decisions:** List all definitive decisions made during the meeting. Each decision should be a concise statement. If no decisions were made, state "None". 2. **Identify Action Items:** List all tasks assigned to individuals or teams. For each action item, include: * **Action:** A clear description of the task. * **Assignee:** The person or team responsible. * **Due Date (if specified):** The deadline for completion. If no due date, state "N/A". 3. **Summarize Discussion Points:** Extract the main topics discussed and the key takeaways or conclusions from each. Avoid including irrelevant chit-chat. Present each point concisely. 4. **Identify Next Meeting Details (if specified):** If mentioned, extract the proposed date, time, and topic for the next meeting. If not specified, state "None". **Meeting Notes:** [MEETING_NOTES_TEXT] **Output Format (JSON):** ```json { "decisions": [ "Decision 1", "Decision 2" ], "action_items": [ { "action": "Task description", "assignee": "Name/Team", "due_date": "YYYY-MM-DD or N/A" } ], "discussion_points": [ "Summary of discussion point 1", "Summary of discussion point 2" ], "next_meeting": { "date": "YYYY-MM-DD or N/A", "time": "HH:MM or N/A", "topic": "Topic or N/A" } } ``` If a category is empty, ensure the corresponding array is empty (e.g., `"decisions": []`). Ensure all fields are present in the output JSON. Prioritize clarity and conciseness.
Structured, task-focused, reduced hallucinations

Engineering Rationale

The optimized prompt works better for Qwen 2.5 72B because it provides an explicit, step-by-step chain of thought, guiding the model through the extraction process. It defines clear categories (decisions, action items, discussion points, next meeting details) and specifies the exact structure for each, including sub-fields for action items. Crucially, it dictates a strict JSON output format, which significantly improves parsing and reduces ambiguity. This structured approach leverages the model's ability to follow complex instructions and adhere to formatting rules, leading to more accurate, consistent, and machine-readable output. The 'vibe_prompt' is too vague and open-ended, leaving too much interpretation to the model, which can result in inconsistent formatting and missing information.

0%
Token Efficiency Gain
The output JSON should strictly adhere to the specified schema, including handling of empty arrays for missing categories.
All action items must include 'action', 'assignee', and 'due_date' fields.
Decisions and discussion points should be concise and directly reflect the meeting content.

Ready to stop burning tokens?

Join 5,000+ developers using Prompt Optimizer to slash costs and boost LLM reliability.

Optimize My Prompts