FRB Collect
Privacy notice
Updated 6 October 2026
This notice describes how FRB Collect processes personal data when pilot companies use the service for real orders. Controller responsibilities, purposes, retention and deletion are agreed in writing with each company before the pilot begins.
Who provides FRB Collect?
The service is provided by FRB Collect Crisera-Eilertsen, Norwegian registration number 938 572 526, Mettedalsfjellet 27, 5302 Strusshamn, Norway. Privacy contact: kontakt@frbcollect.com.
FRB Collect is the controller for its own contact and account relationships. When we process a customer’s order data under that customer’s instructions, FRB Collect is the processor and the customer is the controller. This processing is governed by a data processing agreement.
When an agreement is accepted electronically, we store the complete agreement text, version, company details, signed-in user identity, time and confirmed authority. The receipt can be downloaded in the portal and is removed from the active customer database with the account no later than 30 days after effective closure or a confirmed deletion instruction. Separate retention only takes place where a specific legal obligation or legal claim requires it.
Pilot enquiries from the website
When you request a pilot call on the website, we send your company name, email address and preferred meeting time to FRB’s contact mailbox at Namecheap to respond to your enquiry. This does not create an account, pilot agreement or subscription. The submitted form content is not stored in the application database. Technical keyed fingerprints and delivery status are used to limit abuse and duplicate messages; they are removed from the active database after one day at the next maintenance run. The enquiry in our contact mailbox follows the retention period for contact and support enquiries below.
Why we process personal data
For contacts and users at business customers, we rely on our legitimate interest in managing the customer relationship and securing access (GDPR Article 6(1)(f)). We limit this to necessary contact and security information. You may object to this processing. If you are personally a party to a contract, we process necessary data to enter into or perform it (Article 6(1)(b)). Accounting data, if any, is processed under applicable legal requirements. The customer determines the lawful basis for order data that we process on its behalf.
Contact details come from you or the company that grants you access. Order data comes from company uploads or its selected inbox, folder or label. Necessary account information must be provided so that we can create and manage access. FRB does not make solely automated decisions with legal or similarly significant effects on individuals; orders require human review.
Trial without an account
The public trial must be used only with fictional or anonymised information. Order text, products and trial mappings are sent to FRB’s server to produce a result. The trial does not create a customer account or order and does not store the content in the customer database. Trial mappings exist only in the open tab and disappear on reload. Ordinary technical network traffic may be logged by the hosting provider.
What the service processes
FRB Collect processes account information, product catalogues, name mappings, order drafts, review notes and technical activity needed to provide the service. When an email account is connected, we also store an account identifier, the selected folder or label, encrypted access tokens or IMAP credentials, and identifiers for messages already processed.
Gmail data and purpose
The Gmail connection requests read access. It reads messages in the inbox or label selected by the user, extracts text and supported PDF attachments, identifies possible orders, creates order drafts and prevents duplicate imports. FRB Collect does not send, change or delete email.
Google permission technically covers the mailbox, although FRB Collect limits intake to the selected inbox or label. Inbox reading starts only when enabled by the user. A dedicated order mailbox can limit the information FRB needs to read.
Automatic inbox sorting
Automatic inbox sorting also reads messages that may turn out to be ordinary correspondence, to determine whether they contain an order. Sorting runs locally on FRB servers without an external AI service. Content and attachments of skipped messages are not stored in the customer database; a message identifier and processing status are retained to prevent reprocessing. Uncertain possible orders are saved as drafts for review under the normal retention periods. Original messages are not changed or deleted. Sorting uses Norwegian and English rules and can misclassify messages.
Microsoft 365 and other email providers
The Microsoft connection uses delegated read access to the user’s mailbox or read access to a shared mailbox when selected. The signed-in Microsoft user must already have access to the shared mailbox. Microsoft’s permission may cover more than the selected folder; FRB limits collection to the folder chosen by the customer.
For supported IMAP providers, FRB stores the server name, username and mailbox password or app password in encrypted form so that collection can run automatically. FRB must be able to decrypt this password; unlike a portal sign-in password, it is not a one-way hash. The connection uses TLS with certificate verification. FRB reads without marking messages as read or sending, moving or deleting them. The provider password may nonetheless grant broader technical access. Use a dedicated order mailbox and a separate app password where supported.
Use, sharing and sale
Information obtained from Google APIs is used only for the order processing shown in the service and for necessary security, operations and troubleshooting. It is not used for advertising, credit scoring or training general-purpose AI models. FRB Collect does not sell Google user data. Data is not shared with other customers. Necessary service providers may process data on our behalf for agreed purposes.
FRB Collect’s use and transfer of information received from Google APIs complies with the Google API Services User Data Policy, including its Limited Use requirements. This describes our commitment on data use; it does not mean that Google has approved the app.
Text recognition in PDFs
Scanned PDF pages may undergo text recognition on FRB’s existing Railway server. Documents are not sent to OpenAI or another external AI service for this reading. Temporary PDFs and page images are used during processing. Time and size limits apply, and temporary files are cleaned up on normal completion or handled errors. After an abrupt server stop, remnants may remain until the runtime is cleaned up; this must be followed up as an operational error. Extracted text and order drafts follow the retention rules below. Text recognition may misread characters and numbers, so its output is flagged for review.
Retention and security
The service runs on Railway with web, worker and production database services in Amsterdam, the Netherlands. On 6 October 2026, Railway confirmed that services and attached volumes in EU West, and native volume backups, are stored in the deployment region. This does not confirm the location of older backups from previous regions. Railway application, deployment, build and HTTP logs are stored in US West, USA, for 90 days from each log entry. The Pro plan’s 30-day viewing window is only a query limit, not a deletion deadline. Railway states that these logs cannot be deleted early or selectively. These are technical provider logs, separate from orders and activity history in FRB’s customer database. FRB must minimise personal data in logs and must not log order content, passwords or access tokens. Necessary support access may originate from countries where Railway’s global team works; Railway has not supplied a complete country list. Access controls, relevant subprocessors and transfer mechanisms remain under clarification and must be documented before a pilot starts with real customer data.
The website uses HTTPS. The separate field containing extracted source text and access tokens is encrypted by the application; passwords are stored as one-way password hashes.
An order review in progress is stored as a separate encrypted draft for the signed-in user. The draft is deleted on approval or rejection, or after seven days without changes during the next maintenance run. It does not approve the order. Personal work views and the chosen start page are stored encrypted for each user until the view or account is deleted. Undo history for product changes and deleted name mappings is encrypted separately, remains available for seven days and is deleted in the next maintenance run after that period. This history is also deleted when the company account is deleted.
Source text is used to review and correct order drafts. It is hidden 30 days after order creation, even if maintenance is delayed. The background process deletes the source field after 30 days during the next successful maintenance run. Generating a CSV does not automatically delete order data or review history. The CSV is generated in memory and is not stored as a separate file on FRB’s server. Original emails remain in the customer’s mailbox; FRB does not permanently store original attachments.
Raw source copies in order lines are also deleted after 30 days. The subject, structured order lines, warnings, review notes and other retained order information follow the separate order retention period. This order data is retained for 365 days from receipt (a 12-month retention period), and activity for 365 days from the event. Corrections or a subscription change do not extend the period. Expired orders are hidden from the portal and exports and deleted at the next successful maintenance run. Not every database field is encrypted at application level. Backups follow the rules below.
Termination and backups
On effective termination or a confirmed deletion instruction, we stop access and collection and delete the company’s data from the active database within 30 days at the latest. Any return of data is agreed and carried out before deletion. Cancelling a future paid agreement must not remove access before the paid period ends. An agreed transition from pilot to paid service retains the same account and setup under the usual retention periods.
FRB’s selected schedule retains daily backups for 6 days and weekly backups for 27 days. On 6 October 2026, Railway confirmed that backups are deleted at expiry or when manually deleted, with no subsequent recovery window. Under this schedule, deleted information may remain in backups for up to 27 days after active deletion, normally up to 57 days after effective termination. Railway monthly backups have 89-day retention but are not part of FRB’s selected schedule. Manual backups have no automatic expiry and remain until deleted; Railway cannot add automatic expiry to an existing manual backup. An older manual backup is currently retained by the owner’s decision. Its contents and processing location, as well as old volumes and backups from previous regions, must be clarified before the pilot agreement is concluded. Manual backups containing customer data require a documented deletion deadline, a responsible operator and verified deletion. Backups are used only for recovery. Previous deletion instructions must be reapplied before customer access is reopened after recovery. These backup deadlines do not apply to Railway technical logs, which have the separate 90-day retention described above.
Product catalogues, saved name mappings, column layouts and account settings are kept while the account is active or until removed. Minimal message identifiers are kept until the account is deleted to prevent importing old messages again; these identifiers can still be linked to the mailbox. Contact and support cases are reviewed manually and deleted no later than 12 months after closure. Unnecessary attachments are removed earlier. Contract and accounting records may be kept separately where a specific legal obligation or legal claim requires it.
Who can access the data?
Account users can access their own company’s information. FRB’s operations view shows technical status and does not provide access to other companies’ orders. Infrastructure administrators may nonetheless have broader technical access. Routine troubleshooting should first use technical status and synthetic examples. Reading a specific message or order for support requires a defined case and the customer’s express confirmation. Necessary security work and legally required access must be justified and documented separately.
Cookies and help
Sign-in uses the cookie frb_session for up to 12 hours. It is deleted on sign-out. FRB has no advertising or analytics scripts in the portal. The helper displays prewritten answers locally and does not send questions or orders to an external AI service.
Saved column layouts
Confirmed column layouts are stored with the sender address, column headings and selected meanings in the company account. A layout is reused only for the same sender in the same company. It can be removed under Products and names and is included when data is returned or the company account deleted. Confirmation is recorded in activity and in each order’s time-limited review history.
Review history
For new orders, we store the initial order draft and the outcome of approval or rejection, with a review note and any name mappings the user chooses to remember. This history is encrypted separately and available to the company only for 30 days after the order was received. After that, it is hidden and deleted at the next successful maintenance run. The history supports traceability and quality review; it is not used to train external AI models. Approved orders and saved name mappings follow the company’s agreed retention and deletion procedures. Backups have separate retention periods.
Access, disconnection and deletion
The company can pause collection or disconnect the account in the portal. Disconnection deletes stored access tokens and IMAP credentials from FRB’s active database. Orders already imported follow the retention periods above. The customer should also revoke the app’s access with Google or Microsoft, or revoke the IMAP app password with the email provider. Backups follow the periods above. Requests for access, correction, export or deletion should be sent to kontakt@frbcollect.com. We verify identity and company affiliation before carrying out such requests.
Disconnection alone does not delete previously imported orders. The customer manages downloaded CSV files and original messages in its mailbox. When FRB processes orders on a company’s behalf, we assist the company with data subject requests. Depending on the processing, you may also have rights to restriction, objection and data portability. You can complain to the Norwegian Data Protection Authority.
For data for which FRB is the controller, we respond to rights requests without undue delay and within one month at the latest. Where necessary because of complexity or the number of requests, this may be extended by up to two months; we notify you of the extension and reason within the first month. Requests are generally handled free of charge. If a request is refused, we explain why and describe complaint options. For a customer’s order data, we assist the controller customer in meeting the applicable deadlines.
Payment and subscriptions
When you choose a paid agreement, FRB Collect uses Stripe for payment, invoicing and subscription management. Stripe receives the information you enter at checkout, such as name, billing address, email address and payment details. FRB stores the Stripe customer reference, the terms you accepted, subscription status and necessary invoice and payment references. Card numbers and security codes are not stored in the FRB database. Order contents, PDFs and customer mailbox credentials are not sent to Stripe.
Payment information is used to perform the agreement and manage payments. Information that must be retained under accounting rules or to handle legal claims is subject to separate retention periods and is not automatically deleted with order data. Stripe describes its processing and any international transfers in its privacy policy.
Service providers
Railway hosts the application and database. Namecheap provides the domain and Private Email for kontakt@frbcollect.com and support@frbcollect.com. The customer’s chosen Gmail, Microsoft 365 or supported IMAP mailbox is a source connected by the customer. HetrixTools monitors availability through a technical endpoint that shows only whether the service is available. The endpoint does not send orders, email contents, customer names or customer identifiers. HetrixTools processes FRB account and alert information; operational alerts go to FRB’s Gmail account. We do not use an external AI provider to read orders. Do not send passwords, recovery codes or complete customer orders in ordinary support email.
Processing outside the EEA requires a valid transfer mechanism and necessary safeguards. Before a pilot begins, we document relevant recipients, processing locations and transfer bases in the provider annex. Contact us for more information and copies of relevant safeguards. A public privacy notice does not replace these checks.
Contact and changes
Privacy or security questions can be sent to kontakt@frbcollect.com. Material changes to data use will be described before the new use begins. This privacy notice provides information about the processing. Confirming that it has been read does not constitute general consent to use data for new purposes. The customer’s data processing agreement governs our processing of order data on its behalf.