Import a general ledger into an analysis project

Overview

An analysis project reads a client's general ledger through a profile. A profile says how one kind of export is written: which column holds the account, the entry date or the amounts, how dates and numbers are written, and which texts mean debit or credit. Blast Audit provides profiles for several official formats. For any other export, you map its columns once, or let AI propose them, and save the result as a profile for your firm.

If the Overview of your project shows no Profile list, ledger profiles are not switched on for your organization yet, and the project imports a FEC as before.

Formats read today

Blast Audit provides a profile for each of these exports:

Country: France · Export: FEC

Country: Lithuania · Export: i.SAF-T 2.01

Country: Norway · Export: SAF-T Financial 1.30 and 1.40

Country: Poland · Export: JPK_KR_PD

Country: Portugal · Export: SAF-T (PT) 1.04_01

Country: Romania · Export: SAF-T D406

Country: Netherlands · Export: XAF 3.2 (XML Auditfile Financieel)

Country: Any · Export: QuickBooks Online General Ledger report, exported as CSV with its English column names

The Profile list may show other profiles. It puts first, under Matches this file, the profiles that recognize the file from its header.

For any other export, you choose the column of each field yourself (Map the columns) or let AI propose them (Propose with AI). This works for:

a CSV or text file whose first line holds the column names, split by a tab, a semicolon, a comma or a pipe;

an Excel workbook (.xlsx or .xlsm): choose the sheet that holds the ledger.

For a GoBD export, add the index and its table files, their folder, or the export's ZIP archive. If no profile in the list reads it, choose the table that holds the ledger and map its columns. AI does not propose a mapping for a GoBD export. An XML file can be read only through one of the profiles above.

Import a ledger

You need full access to the project.

1. Open the project, then Data, and add the file.
2. Open Overview. Under Profile, choose the profile that reads the file, or click Map the columns or Propose with AI.
3. If the profile leaves a choice to you, such as the encoding or the kind of ledger, make it.
4. Click Check the file. The check reads the whole file and shows the number of lines and entries, the total debit and total credit and whether they are equal, the dates, what the import found, and the first lines as the import reads them.
5. For a mapping you made or AI proposed, compare these lines and totals with the file, then click Validate. If a column is wrong, change it and check again.
6. Click Launch the tests.

Tests that do not run

Some tests need a column the file may not have, such as the user who posted an entry or the time it was entered. Before the launch, the screen lists the tests that will not run on this file and why. The results then show them as Not run, with the reason: a missing column, times not in local time, or a setting left empty.

Firm profiles

After you validate a mapping you made or AI proposed, you can keep it for your firm: type a Profile name and click Save for the firm. Any member with full access to the project can save one.

Every member of your organization then finds it in the Profile list as "name, firm profile". When a file has the columns the profile reads, the list offers it under Matches this file. Each file is still checked before its tests run.

The dashboard does not yet let you rename, change or remove a firm profile.

What a firm profile keeps, and for how long

A firm profile holds no line of the ledger. It holds what reading the file needs, copied from the file you mapped: the column names, how dates and amounts are written, the texts that mean debit or credit, and the name you gave it, with who saved it and when.

A firm profile belongs to your organization, not to a project: closing or deleting a project does not delete it. It is kept while your contract runs, and deleted with your organization's other data when the contract ends, as the Data Processing Agreement sets out. The journal of each run keeps a copy of the profile the file was read with, so the run replays the same way if the profile changes later. That copy is deleted with the project.

What AI receives when it proposes a mapping

Propose with AI works on a CSV, text or workbook file. Blast Audit's code first reads the file's encoding, separator, date formats and number formats itself. AI is then asked which field each column holds. It receives:

the name of each column, and its values in at most 20 of the first lines under the header, empty values left out, each name and value cut at 200 characters;

when the amounts are in one column with a sign, a second question on the same lines: the account, the account name, the description and the amount of each line, to tell whether a positive amount is a debit or a credit.

A file with more than 64 columns is not sent. AI receives neither the file name, nor the project or client name, nor your organization's name. The requests go to Azure OpenAI in the EU Data Zone. Microsoft temporarily stores the requests sent to Azure OpenAI and its responses for abuse monitoring; those copies are not deleted with the project (see Where your data is stored and for how long).

The fields AI filled are marked AI. A field it was less sure of is left to you. Nothing is imported or saved until you check the file and validate it. A proposal uses your organization's AI credits.