3 min read
What makes your records useful for AI?
Look for a request, the work, and the result.
Look for a sequence of work
Look for what was asked, what your team did, and what happened. Together, these parts make the work easier to understand.
A support case might show the problem, checks, and confirmed fix. Keep these parts together.
Keep the context that explains decisions
A final answer may not explain why it was chosen. Useful details include roles, dates, task status, and related records.
Keep only the details the agreed task needs. Review private information before sharing.
Describe quality before counting volume
Count complete tasks as well as messages or rows. Note duplicates, missing results, automatic replies, and old instructions.
Explain what your count means. A thousand messages are not the same as a thousand resolved cases.
Confirm the right to use the source
Check who owns the content and what sharing rights apply. Customer terms, employee information, and material from others can restrict use.
List the sources and open permission questions before preparing a sample.
See what your data could earn.
Describe your records to Korra and get a starting range. Your files stay in your systems.
What’s my data worth?