Lesson 6: Few-Shot Learning
See how adding examples (few-shot) dramatically improves AI consistency and quality

Extract action items from meeting notes with consistent formatting

📚 Available Examples

Example 1:

Input: John said we need to finalize the speaker list by next Friday and get the contracts signed.

Output: **Action Item**: Finalize speaker list and obtain signed contracts **Owner**: John **Deadline**: Next Friday

Example 2:

Input: Sarah mentioned she'll reach out to the venue about AV requirements before the end of the month.

Output: **Action Item**: Contact venue regarding AV requirements **Owner**: Sarah **Deadline**: End of month

Example 3:

Input: The committee agreed that Mike should draft the sponsorship prospectus and circulate it for review by March 15th.

Output: **Action Item**: Draft and circulate sponsorship prospectus for review **Owner**: Mike **Deadline**: March 15th

💡 Key Takeaways

When to Use Few-Shot Learning:

  • Consistency: When you need the same format every time
  • Style matching: When you want a specific tone or voice
  • Complex tasks: When instructions alone aren't enough
  • Edge cases: Show examples of tricky situations

Practice Tip: Start with 2-3 high-quality examples. More isn't always better—focus on diverse, representative examples that cover different scenarios.