Docs

What makes a good report file

Docents looks for structured, tabular data — spreadsheets, tables in a document, or a database schema. It extracts field/column names, infers each one's data type, and writes a plain-language description of what it represents.

A file that's mostly narrative prose — a memo, a letter, a policy document with no tables — won't produce a useful data dictionary, even if it uploads successfully. If a report comes back with very few fields, or fields that don't look right, the most likely reason is that the source document didn't have much structured data in it to begin with.

Accepted formats, size limits, and which formats are available depend on your plan:

.sql files are read as schema (CREATE TABLE statements) — Docents reads the column definitions directly rather than inferring them, and fills in a plain-language meaning for each one. .sql upload is Professional-only, as shown above.

Instead of uploading one file at a time, you can upload a single password-protected .zip containing multiple reports. Each file inside becomes its own report, processed independently:

• The zip must be password-protected with AES-256 encryption (the standard modern option in most zip tools — not the older, weaker "ZipCrypto" method). The password is used once to open the archive and is never stored.
• Maximum zip size: 50MB. Each file inside is still capped at your plan's own per-file size limit (table above).
• A wrong password or a corrupted archive fails the whole batch — nothing inside gets processed, and nothing is charged against your report credits.
• Once the password is correct, each file is processed on its own: one bad or unsupported file inside the zip only fails that file, not the rest of the batch.
• There's no live progress bar for a batch — it can take a while, so feel free to close the tab. You'll get an email with a full summary (what succeeded, what failed and why, and any conflicts found) once every file is done.

Exporting your data dictionary

Every completed report can be exported as:

• CSV — the extracted fields as a flat table (element name, display name, type, meaning, source, formula), on every plan.
• Excel (Starter and Professional) — the same table as an .xlsx file.
• SQL (Professional only) — the same field data as SQL, in two modes:
  • With CREATE TABLE — a full script: CREATE TABLE for a fresh table matching the dictionary's shape, followed by INSERT statements for every field.
  • Insert statements only — just the INSERT statements, for when you already have a table to load the dictionary into.

Note: exports carry the data dictionary Docents extracted — field names, types, and descriptions — not the original report's row data. Docents never stores the contents of your report beyond what's needed to build that dictionary.

Archiving vs. deleting a report

Once your Reports list gets long, you have two ways to clean it up — they do genuinely different things, so pick based on whether you still want that report's fields in your Data Dictionary.

• Archive — removes the report from your list and deletes the uploaded file (you already have your own copy), but keeps every field it contributed in your Data Dictionary. Use this to tidy up a long list without losing what's already been extracted.
• Delete — permanently removes the report and every field it contributed to your Data Dictionary. Use this to undo a report you uploaded by mistake.

Both are permanent — neither can be undone once confirmed.

How your data is handled

Uploaded files and the fields extracted from them are stored in your account, private to your organization. They're processed through our AI provider's API to identify fields — never used to train any AI model, but they are retained until you delete the report or your account. We recommend using anonymized or demo data, or a blank version of your report with just the structure (column headers, field names, no real values), wherever possible. See the Privacy Statement for the full picture.