Back to search
Investments
Azure OpenAI
Research & analysis
Internal

Auto-tag investment research library by theme

PS
Priya Shah · Senior Investment Analyst
Last reviewed 2025-04-20

The problem

Research library has 12K+ documents with inconsistent tagging; analysts can't find prior work.

Business context

Re-doing research that already exists is a hidden cost across the investment org.

The solution

Embeddings + topic model assigns themes; analyst confirms in batches.

Outcome & impact

Time saved
Hard to quantify; high qualitative impact
Outcome
Findability up sharply (analyst survey: 4.6/5 vs 2.1/5 baseline).

Reusable assets

Prompt
Assign up to 3 themes from the controlled vocabulary to this document. Return JSON: {themes: [...], confidence: 0–1}.
  • Theme taxonomy(link)
#investments
#knowledge-management
#tagging