ValueNova Limited engages a small number of third-party service providers ("Sub-processors") to deliver and support the platform. We are committed to transparency about who processes Customer Data on our behalf.

This page lists the Sub-processors we engage. Customers are notified at least thirty (30) days before any new Sub-processor is added or replaced. The full process and customer rights are set out in our Data Processing Agreement.

What this list covers. It lists the third parties that process Customer Data on our behalf: the data our customers put into the ValueNova platform, which we process for them under our Data Processing Addendum. It does not cover processing we carry out for our own purposes, such as website analytics (Google Analytics), marketing and sales. That is described in our Privacy Policy and Cookies Policy.

Current sub-processors

Sub-processor	Service provided	Data processed	Location
Google Cloud Platform (Google Cloud EMEA Limited)	Application hosting (Cloud Run), databases (Cloud SQL), file storage and backups (Cloud Storage), and release builds (Cloud Build)	Hosting of the Service; structured customer data of workspaces held in their own database, and the central account directory; uploaded documents and generated exports; nightly backups of those databases, kept 35 days; complete copies of the shared production database taken before each release, eligible for deletion 7 days after each is taken; a second copy of the nightly backups in a separate, access-restricted project, kept 34 days	EU — `europe-west1` (Belgium); the second backup copies in `europe-west4` (the Netherlands)
Supabase, Inc.	Managed PostgreSQL database and authentication	Structured customer data (accounts, assessments, value-model data) of workspaces on the shared database; sign-in for every workspace; and, for a workspace moved into its own database, the copy the move leaves behind (never served, kept for a move back, deleted at offboarding)	EU — Stockholm, Sweden
Anthropic, PBC	AI inference (Claude) for value-modelling features	Prompts/inputs derived from assessment data — see "AI and your data" below for exactly what is removed before sending	US
OpenAI OpCo, LLCOpenAI Ireland Limited processes EEA and Swiss personal data, as provided in the OpenAI Data Processing Addendum.	AI embeddings for document search and speech-to-text (Whisper) for voice capture	Uploaded document text; voice-capture audio and transcripts	US
Resend	Transactional email delivery	Names and email addresses of users (invitations, notifications)	US
Sentry (Functional Software, Inc.)	Application error monitoring	Diagnostic/error data (configured not to send IPs or cookies)	EU (Frankfurt)
Upstash, Inc.	Rate-limiting cache	Technical identifiers (IP address / user ID)	EU — `europe-west1` (Belgium)

AI and your data

ValueNova uses third-party AI providers — Anthropic (Claude) and OpenAI — to power features such as value-model generation, document search, and voice capture. When you use these features, the relevant content (for example assessment inputs, uploaded documents, or meeting audio) is sent to these providers solely to return a result to you. Before assessment content is sent for AI generation we remove email addresses, US-format telephone numbers, and payment card numbers written in groups of four. Telephone numbers in other formats, including UK and European numbers, are not removed, and a card number written without spaces or dashes is only partly removed. Names and company names already held on the deal are replaced with tokens; other names in the text are not. That mapping is reversible, so the content remains personal data and we do not describe it as anonymised. Unless your administrator has chosen to allow AI insights that name individual members of staff (off by default), deal-level AI also replaces the full names of your own staff with "a colleague" before the request is sent; that replacement is not reversed.

Where these protections are bypassed or limited. Four uses bypass or limit them, each deliberately:

Prospect research sends the real company name, its domain and the seller's notes. Researching a named company is the whole purpose, so nothing is replaced.
Whole-account questions — the organisation-wide assistants and the Analytics question box — send aggregate report figures together with customer company names. By default they send no names of your own staff and describe your team as a whole. If your administrator chooses to allow AI insights that name individual members of staff, those names are sent too. Staff email addresses are removed before the request is sent in either case.
Turning a voice transcript into structured fields sends the transcript to Anthropic. If the transcript is not linked to a deal, nothing is replaced and it is sent as spoken. If it is linked, only names already held on that deal are replaced, so a person newly named in the conversation is not.
Uploaded documents and voice recordings are sent as provided: document text to OpenAI for search indexing, and audio for transcription. Nothing is removed or replaced on either path.

These providers process your data under their commercial API terms, under which your data is not used to train their models.

We do not have a zero-data-retention agreement with either provider. Anthropic declined our request on 13 August 2026; it offers zero retention only under enterprise agreements with significant usage volume and contract value. Zero data retention and EU regional processing are not available to ValueNova on OpenAI's standard terms at our current stage. We reviewed the commercial route required to obtain them in August 2026 and decided not to proceed at this time. We will revisit this if a customer requires it. Your data is therefore retained by these providers under their standard terms; we do not describe it as deleted immediately. It is not used to train their models.

Administrators choose how AI runs for their organisation: Frontier (Claude/OpenAI — best quality), Local (a model you host yourself), or Off (AI features disabled entirely). The in-app AI Governance Center controls AI usage, capabilities and budgets.

What Local and Off actually change. They change where business-case AI runs. Two capabilities behave differently, and it is worth being precise about them:

Document search runs on OpenAI embeddings and cannot run on a self-hosted model, because a search index is only comparable to vectors produced by the same model. Selecting Local therefore switches document search off rather than re-pointing it. An administrator may enable it for search alone; where they do, uploaded document text is sent to OpenAI.
Voice transcription follows the selection: on Frontier it uses OpenAI Whisper, on Local it uses the speech endpoint you configure, and on Off it is disabled.

Under Anthropic's and OpenAI's standard commercial API terms, your content is not used to train their models.

Customer rights

When ValueNova adds or replaces a Sub-processor, ValueNova will:

notify the Customer at least thirty (30) days before the change takes effect;
allow the Customer thirty (30) days to object in writing on reasonable data protection grounds; and
if the Customer objects and the parties cannot resolve the matter, either party may terminate the affected Order Form on written notice with a pro-rata refund of any prepaid, unused fees.

All Sub-processors are bound by data protection obligations no less protective than ValueNova's own commitments under the customer's Data Processing Agreement.

Data residency

ValueNova hosts customer data in the European Union (Belgium). Application compute runs on Google Cloud Run, structured workspace data and the central account directory are held in Google Cloud SQL, and files are stored in Google Cloud Storage, all in the europe-west1 region; nightly backups of those databases are also kept in Belgium, with a second copy of each backup held in a separate, access-restricted project in europe-west4, the Netherlands. Workspaces still on our shared database, and sign-in for every workspace, are held via Supabase in Stockholm, Sweden (North EU).

We do not currently offer per-customer choice of hosting region. AI inference (Anthropic, OpenAI) is processed in the US — see "AI and your data" above. Customers with specific data-residency requirements should contact — additional regional hosting options for Enterprise customers are on our roadmap.

International transfers

Where personal data is transferred outside the United Kingdom or the European Economic Area to a jurisdiction not deemed adequate under UK GDPR or EU GDPR, ValueNova relies on:

From the UK: the UK International Data Transfer Agreement (IDTA) or the UK Addendum to the EU Standard Contractual Clauses, in each case as issued by the Information Commissioner's Office.
From the EEA: the EU Standard Contractual Clauses (Commission Implementing Decision (EU) 2021/914) Module 2 (controller-to-processor).

A Transfer Impact Assessment is available on request to.

Note on website analytics

ValueNova's marketing website at valuenova.ai uses Google Analytics 4 for aggregate website analytics, configured with IP anonymisation enabled and no advertising features. Google Analytics is a website-only analytics tool and is not a Sub-processor of Customer Data under the Service. See our Cookies Policy for full details.

Contact

For sub-processor questions, objections, or DPA queries:

ValueNova Limited

71-75 Shelton Street

Covent Garden, London WC2H 9JQ