Our sub-processors
Last updated: 15 September 2026
To run the Tupi software we rely on a small number of other companies, for hosting, for sending and receiving email, and for a few specific tasks. In data protection language these are our sub-processors. They only ever handle personal data on our instructions, under a contract that binds them to the same obligations we owe you.
This page lists the ones that matter. We keep it current, and section 6 of our Privacy Notice links here.
Who they are
| Provider | What they do for us | Where it is processed |
|---|---|---|
| Amazon Web Services | Hosting, databases, file storage, sign-in and account security, sending and receiving email, background processing, and the AI processing described in our Privacy Notice (reading uploaded job sheets, receipts and supplier invoices, and classifying interpreter email replies), which runs on Amazon Bedrock. This is where the Software itself runs. | United Kingdom. Inbound email is received in Ireland. |
| Google Workspace. Our company email and shared documents. Anything you email us is held here. | Contracted through Google Cloud EMEA Limited in Ireland. Google may process some of it elsewhere under its data processing terms. | |
| Xero | Our own accounting records, including the billing contact for each Agency we invoice. | United Kingdom |
| Mintly | Validating bank account details, checking that a sort code and account number are valid before a payment is set up. | United Kingdom |
| VoIPstudio | Telephony. It carries on-demand interpreting calls that interpreters take in the browser, and sends the text-message codes used to verify a phone number. | United Kingdom |
Accounting packages an Agency connects
An Agency can connect its own accounting package so that the invoices and interpreter bills raised in the Software appear in its books. The packages the Software connects to are Intuit QuickBooks, Sage Accounting and FreeAgent. QuickBooks is available now; Sage Accounting and FreeAgent are released to agencies as each completes testing. An Agency connects one package at a time.
That package is not our sub-processor. The Agency chooses it, holds the relationship with the provider, and remains responsible for the data sent to it. We send customer and supplier records, invoices and bills on the Agency’s instruction, with their supporting documents (job sheets, receipts and an interpreter’s own invoice) where the package accepts them. The Agency’s own privacy notice governs what happens next.
AI processing, and a provider we do not use in the live service
All AI processing in the live service runs on Amazon Bedrock, within Amazon Web Services, and is covered by the Amazon Web Services entry above. No separate AI provider receives your data.
We do use Anthropic in our test and development environments, but only ever with non-real test data. No personal data belonging to an Agency, an interpreter or anyone else is ever sent to it. That is why Anthropic is not listed as a sub-processor above.
Services that receive no personal data
We use Google’s mapping services to check that an assignment address is real and to estimate journey time and distance. Those requests carry only an address or a pair of map coordinates, never a name, never a booking reference, and nothing that identifies who the assignment is for. We have assessed that no personal data leaves us through them, which is why Google’s mapping services are not listed as a sub-processor above.
When this list changes
We keep this page up to date. If we add a sub-processor, or change what an existing one does, we update this page and tell the Agencies affected so they have time to consider it.
Questions
Email info@tupi.solutions and we will answer. If you would like the detail of the contracts and transfer safeguards behind any entry above, just ask.