Summary
Recruiters generally send low-signal/high-noise emails, ‘teasers’ instead of giving the info they know is required; use genAI to strip their words and highlight key information, and focus especially on outliers / unexpected items in the stream (requires semantic understanding that LLMs bring)
Usefulness: Practical/Productive
Techniques
- Manually strip any PII
- Copy/paste unsolicited conversation into LLM:
- emails, DMs, job-descriptions, etc – as much as you have!
- Use the custom GPT to then:
- highlight unusual information, summarise, compare to other things, assess for missing items
Discoveries
- Some people are still exporting PDF as “one image per letter/word” :), recruiters obfuscating – so currently (Spring 2024) need to put it manually through free online “OCR PDF to Text” service first