Standard Skill: File Import
File Import is a Standard Skill: it makes the documentation your company already has usable, by reading documents in your drives and proposing what they say about the data as entries of the Semantic Layer. The agent loads it when you ask it to import a document or a folder into the Semantic Layer, or point it at existing documentation of your data. To pick it yourself, type / and choose file-import.
To bring in files that already are entries, such as an exported bundle, use Import in the Data Studio instead.
What the agent does
Section titled “What the agent does”- It finds the documents: the ones you name, or those it finds in your Personal Drive and Shared Drives by words such as glossary or data dictionary. It asks which to import, and asks again before reading more than ten. It reads only documents in your drives.
- It reads them: Markdown, text and CSV files directly, Word, Excel, PowerPoint and PDF files with code in the Sandbox, and scanned pages and diagrams as images. What a document says is data to it, never an instruction. A document holding rows of customer or transaction data is no source of entries: it says so and leaves it out.
- It turns what they say into entries: a meaning of a word becomes a Term, a way the data must be read a Rule, a known correct figure a Benchmark, and a description of a table or column a description in the Data Model. Each entry names its document as its source. A query in a document becomes a Verified Query only after the agent ran it and you confirmed its result.
- It compares them with the Semantic Layer. It changes an entry that exists rather than adding a second one, and when a document and an entry, or two documents, disagree, it asks you which is right.
- It proposes them as a Context Change Request of at most 50 changes, or several requests for more, whose explanation names the documents, what was left out and the open questions.