Transcript
Every document you upload becomes structured data. Analytics Manager is where you read it.
From documents to data
DocReveal reads invoices, orders and delivery notes, and writes every one of them into the same shape. Analytics Manager is where that result becomes a table.
Six tabs, one workflow
Six tabs carry the whole flow. Query History keeps every run for audit. Data Catalog documents every column. We start where the work starts.
Datasets and Auditsets
Two checkboxes split the catalog. Datasets hold the business fields themselves: the invoice, the order, the delivery note. Auditsets hold the processing history behind them, and every audit table shares the same structure.
Choosing a dataset
Open the list and you see what DocReveal has already extracted: bank statements, credit notes, delivery notes, purchase orders, invoices. Twenty-four in all. Take the invoices.
Structure, not a flat export
Canonical Invoice Data has thirty-three columns. Some hold one value. Others hold a whole entity — parties, items, sections — where a single column carries several records, each with its own attributes.
Finding the columns you need
You rarely need all thirty-three. Type a word and the list narrows. Type document, and the document fields come up. Type dates, and every date the invoice carries appears. Tick what you want. The counter keeps score, so you always know how wide the result will be.
Run the query
Five columns out of thirty-three. Run the query. What comes back is a clean data table — built from documents that arrived as PDFs, in whatever layout each supplier uses.
Room to read
Two controls give the data room. Expand the result to the full window. Or fold the query panel away to a rail and bring it back when you need it.
Six rows, one problem
Six invoices. Look at the last two rows: same number, same date, same amount. The same invoice reached you twice — and nobody opened a PDF to find it.
Filters without SQL
Prove it with a filter. Choose the field, choose the operator — equals, contains, starts with, is empty — and give it a value. No SQL is written in your browser. The query leaves as a declarative payload, and the tenant filter is added on the server, where you cannot get it wrong. Run it. Two rows. Confirmed.
Save the question
A question worth asking once is worth asking again. Name the view and save it. It comes back with the dataset, the columns and the filter intact — kept private to you, or shared with everyone in your tenant.
Clearing the filter
Delete the filter and the full set is back. Nothing in your data changed — Analytics Manager only reads.
Governed exports
When the result has to leave the screen, Exports writes the same authorized query to CSV or JSON. Same rules, same limits, and still only your own tenant database — your space, separate from everyone else's.
Ask in plain language
And the newest way in: ask in plain language. NL2SQL Chat turns your question into SQL and shows you the query before it runs. This one is brand new — a first version, shipped this month.
The audit layer
Now switch to Auditsets and open Canonical Order Audit. Thirteen columns instead of thirty-three — and the same thirteen on every audit table in the catalog. One shape for all of them, so an auditor learns it once.
Nested values
Run it. Processed at, year, month, day — and the payload itself. One cell, fifteen nested values: the record exactly as it was written, kept beside the run that produced it. That is evidence, one click away.
Your documents, as data you can question. That is DocReveal.