Where your data is stored and for how long
What stays in your workbook, what reaches our servers in the EEA, how long document content is kept, and where to find our policies.
This page sums up what the Privacy Policy and the Data Processing Agreement say about where your data goes and how long it stays. Those two documents are the reference if anything here differs.
In your workbook
The documents you import and the snips you make are saved inside your Excel workbook. They stay wherever you store that file: your computer, a file server, SharePoint or OneDrive.
On our servers, while a task runs
Some features need our servers: OCR, and the automated analysis features. For those, the snipped images, the documents you submit and their metadata are sent to our servers over an encrypted connection and processed there.
That content is held only in short-lived processing contexts:
| Processing | How long the content is kept |
|---|---|
| OCR, document analysis (Probe, AI Extraction) | At most 24 hours after upload |
| Excel agent | At most 24 hours after it reaches our servers; database backups keep it up to 7 more days |
An automated purge deletes expired contexts. If a deletion fails, it is retried on a later pass, so deletion can happen shortly after the deadline. Apart from temporary technical copies (caching, operational logs), which are purged regularly, we keep the content of your documents only where Annex 1 of the DPA lists it: engagements and analysis projects (below), but also financial-statement reviews and PBC folders; in backups, for up to 7 days; and in the copies Microsoft stores for abuse monitoring (next paragraph).
The 24 hours apply to the copies we control. For abuse monitoring, Microsoft temporarily stores all requests sent to Azure OpenAI and all responses; they are not covered by that deadline, and Microsoft's contract sets no duration for them. Only data that triggered Microsoft's automated systems may be reviewed by authorised Microsoft employees.
Our operational logs and error reports are designed not to contain document content.
Matching and AI arbitration
Matching runs on your machine. If your organisation turns on AI arbitration, the rows the engine cannot settle are sent for a judgment, with the document lines around each candidate, and kept only for that request.
Engagements (engagement agent)
If your organisation uses the engagement agent, the documents you import into an engagement, the task journal and the workpapers are kept on our servers while the engagement is open, so the agent can work without Excel open. They are deleted 30 days after the engagement is closed. An engagement inactive for 6 months is closed after a 30-day notice. Before it is closed, you can export the engagement.
Analysis projects and firm formats
If your organisation uses analysis projects, the data sources imported into a project, the datasets built from them, the journal and the results are kept on our servers while the project is open, and deleted between 29 and 30 days after it is closed.
A firm format says how to read one kind of ledger export. A member saves it for the whole organisation, and it is not part of a project: it holds column names and reading settings copied from the file, never the ledger's lines, and it is kept until your contract ends. When a member asks AI to propose a mapping, the column names and at most 20 values per column are sent to Azure OpenAI. Import a general ledger says what a firm format keeps and what AI receives.
Support and Moulinet
Support conversations and their attachments are stored on our servers, on Microsoft Azure in the European Union. When it is active, Moulinet, an artificial intelligence (AI) assistant, receives your messages first and answers only from our public documentation and notes from our team; it hands to the team, without answering, what it recognises as a billing or sales question, or as a data protection or security request that needs a commitment, a document or an action from the team (for example a data processing agreement, a security questionnaire or a request to exercise your rights) or that concerns your own data or account rather than how we handle data and security in general, and anything its sources do not answer. Each of its messages says it is written by an AI. A button in the chat hands the conversation to the team at any time, and Moulinet writes no more in it. A reply to its e-mail does the same, except a reply to a question Moulinet asked to clarify the request: Moulinet may answer that reply once, and the next reply goes to the team.
To handle a message, our servers send the conversation's messages to Azure OpenAI, which sorts the request and writes the reply. The Azure AI service in the EU Data Zone may also check the reply before it is sent. Before they leave, every e-mail address and the phone numbers our tool recognises are replaced; a name, a company or a number it does not recognise may be sent. No data from your account (name, e-mail address, organisation) and no attachment, not even its name, is sent. For abuse monitoring, Microsoft temporarily stores all requests sent to Azure OpenAI and all responses, and may store the requests sent to the Azure AI service and its responses. Microsoft has confirmed to us in writing that, for the Azure AI service in the EU Data Zone, requests and responses are processed within the EU Data Boundary and stored data, including abuse-monitoring data, stays in Sweden, subject to the limited transfers described in its documentation on that boundary; the Microsoft employees who review its data for abuse monitoring are located in the European Economic Area. When Moulinet is not active, no support message is sent to these two services.
Where the servers are
Our infrastructure runs on Microsoft Azure, with production hosting in the European Economic Area, mainly in the France Central region. Where data is transferred outside the EEA, it is covered by safeguards such as the European Commission's Standard Contractual Clauses. Microsoft personnel located outside the EU Data Boundary may also access its processing systems remotely, as necessary to operate, troubleshoot, support and secure its services, under the Standard Contractual Clauses and the EU-U.S. Data Privacy Framework.
The companies that process data for us are listed, with where they process it, in Annex 3 of the Data Processing Agreement. For a security or privacy review, see also our Trust Center.
Account and billing data
To run your account we keep your name, email address, organisation and subscription details, and usage data such as the number of documents and pages processed. Payments are handled by Stripe; we do not store your full card number. We keep this data as long as we need it to provide the service and to meet billing, accounting and legal obligations.
Security and certifications
Data is encrypted in transit (TLS 1.2 or later) and at rest, and each customer is logically isolated. Blast Audit has a SOC 2 Type 2 report issued by Sensiba and an ISO 27001:2022 certificate. More detail is on our Trust Center.
To report a vulnerability, see the Security page.
Your rights
You can ask to access, correct or delete your personal data, or object to its processing, by writing to privacy@blast-audit.com. When a request concerns data in documents that an organisation uploaded, that organisation decides what is done with it and we act on its instructions: ask the organisation first (the firm that uploaded the document), and we help it answer. Our copies of document content are deleted at the deadlines above (database backups keep them up to 7 days longer). The copies Microsoft stores for abuse monitoring are subject to Microsoft’s terms.