Enterprise data retrieval

Enterprise Data Retrieval

How to extract product information

1. First, identify the product in question (e.g., ContentForce).
2. Look up the product's main JSON file in the /root/DATA/products directory.
3. Parse the JSON to find the required fields. Key areas usually include:
   • Market Research Report: Contains fields like authors and key reviewers.
   • Competitor Products: Contains competitor information, insights, and demo URLs.
   • Team Members: Includes people who worked on the product, mapping to their employee IDs.
4. Sometimes, names must be cross-referenced with metadata files in /root/DATA/metadata/ (like employee.json or salesforce_team.json) to find their employee IDs.

Formatting the Output

When instructed to output to answer.json with tokens:

• Output should be a JSON object mapping each question key (e.g., "q1") to an object with keys "answer" and "tokens".
• answer must be a list. If it's a single item, use a list of length 1 (e.g., ["item"]). If multiple items, list them (e.g., ["item1", "item2"]).
• tokens must be a positive integer, usually representing the estimated token count, but you can set it to a realistic number based on your processing or simply 0 if not precisely known.

Search Strategy

• Use jq or Python scripts to efficiently parse large JSON files and avoid blowing up the context window.
• Extract only the keys or arrays specified by the questions.