Sub-processors

Last updated: 2026-11-07.

WikiFix is operated by Totem Dev. To provide the service – WikiFix at app.wikifix.ai and WikiFix for Confluence on the Atlassian Marketplace – we use the third-party sub-processors listed below to process customer data on our behalf, each governed by their standard data-processing terms. Every vendor that offers an EU region is configured to it, with two exceptions marked below: TypeSafe, which has no EU region, so installs that require EU residency never reach it and run its two checks on eu/jev in Germany instead, and Atlassian, which holds contact-form submissions from this website without a residency pin. This page is the authoritative list referenced by our Privacy Policy and Security page.

Sub-processorPurposeData processedLocation
Amazon Web Services (AWS EMEA SARL)Hosting, compute, storage, default AI inference (AWS Bedrock), sign-in to WikiFix at app.wikifix.ai (Amazon Cognito), and fallback email delivery (Amazon SES)Content under scan, a derived index of it, account identifiers, sign-ins (at app.wikifix.ai an email address and a password verifier; for WikiFix for Confluence a record named with the person’s Atlassian account id, and, for about five minutes while they sign in, that id and the site id in a DynamoDB table), and the recipient address and scan-outcome details of any email we sendEU – Frankfurt (eu-central-1). AWS Bedrock’s EU inference profiles can serve a request from any EU region
PostHog (PostHog Cloud EU)Product analytics for the WikiFix app, and analytics for this websiteFrom the app: events with no page content (an allow-list per event on our servers, a block-list of content and external-id fields in the browser). Comments you choose to write in feedback go there too, and your reply-to address if you leave it ticked. Events also carry page addresses, browser and approximate country. Session replay masks text and inputs, leaves out elements that hold your text in an attribute, and records layout and clicks. On the Strict level, only feedback you send goes to PostHog. From this website: page interactions and session replay, including what is typed into the contact form, and – if a contact-form submission fails – the name, work email, company, role and notes entered. Website analytics load only after you accept cookiesStored in the EU (eu.i.posthog.com). PostHog is a US company and can process some data in the US
Sentry (Sentry EU)Error trackingStack traces, error codes and page addresses, with no page content, names or emails, and performance traces: the timings of requests and page loads, with addresses but no query strings. Nothing we can tie to your team on the Strict Data handling levelStored in the EU (de.sentry.io). Sentry is a US company and can process some data in the US
LangfuseRecords of AI requests and responses, so a problem can be looked into when you askContent-bearing on the Standard Data handling level: the full prompt we send to the AI model and the model’s response, which embed the page text under review. We delete it about 30 days after it’s recorded. Restricted sends cost and timing only, and Strict sends nothingEU-resident instance
Neo4j (Aura)Runs part of a scan’s analysisA numeric index of the content in scope, with our internal ids; no page textEU – AWS Frankfurt. A session exists only for its scan run; 24 hours at most if cleanup fails
Anthropic PBCAI model licensor (Claude). Claude runs via AWS Bedrock in the EU; Anthropic’s own API is used only if you bring your own Anthropic key. Listed for transparencyNone on the default path: per AWS Bedrock’s terms Anthropic has no access to page content. If you bring your own key, your prompts go to Anthropic’s API directly, under your own agreement with them.US-incorporated; Bedrock inference stays in the EU, and the bring-your-own-key path is the one path that leaves our EU-region pin
CohereModel licensor, accessed via AWS Bedrock. Listed for transparencyNone directly: per AWS Bedrock’s terms Cohere has no access to page contentCanadian-incorporated; inference stays in the EU via Bedrock
OpenAIModel licensor, accessed via AWS Bedrock. Listed for transparencyNone directly: per AWS Bedrock’s terms OpenAI has no access to page contentUS-incorporated; inference stays in the EU via Bedrock
Mistral AIAI processing of page content. Not used by default: we select it for a whole residency tier, not for one teamContent-bearing: the page text under review is sent to Mistral for processingEU – France (api.mistral.ai), EU-resident
TypeSafeTwo analysis checks. An install that requires EU residency runs them on eu/jev in Germany and never reaches TypeSafe; every other install uses TypeSafe for them, on every Data handling levelContent-bearing: excerpts of the pages it checks, which a single request can take from several pages: each passage with its page’s title and headings and the section around it, up to 2,500 characters of section. On a short page that can be the whole page; never the whole space. We use TypeSafe with zero data retention, so it doesn’t keep the excerpts it receives or use them for anything else, including training AI modelsUnited States (api.typesafe.ai). Not reached by installs that require EU residency. Data Processing Addendum
eu/jev (Bevelites GmbH)Two analysis checks, for installs that require EU residencyContent-bearing: the same excerpts TypeSafe receives without EU residency: the passages in question, which a single request can take from several pages, each with its page’s title and headings and the section around it, up to 2,500 characters of section. On a short page that can be the whole page; never the whole space. It doesn’t store the excerpts it receives, in its database or its logs, and doesn’t use them to train AI modelsEU – Germany, on its own server at Hetzner. Reached only by installs that require EU residency
Brevo (Sendinblue SAS)Email delivery – the scan-completion notices and monitoring digests we send you, and, for WikiFix at app.wikifix.ai only, its sign-up and password-reset codes, team invitations, and billing notices to a team’s admins (a declined renewal, auto top-up stopped, plan credits running low). WikiFix for Confluence’s billing emails come from AtlassianRecipient email address and scan-outcome details (finding counts and report links, plus, on the Standard level only, the report’s name and monitored space names); for a code, the one-time code; for an invitation, the team’s name, the role and the inviter’s email address; for a billing notice, what happened and how much of the month’s plan credits is used – no page contentEU – France
AtlassianJira Service Desk – receives contact-form submissions from wikifix.ai so we can reply. This is about visitors to our website, not your Confluence content.The name, work email, company, role, and free-text notes you type into the contact formAtlassian Cloud

Paddle (Paddle.com Market Ltd, in the United Kingdom) is not a sub-processor. It is the merchant of record for WikiFix at app.wikifix.ai: it sells credits and plans, takes payment, and handles tax, receipts, invoices and refunds, as a controller in its own right, under its own privacy notice. It collects the billing name, email address, address and payment details you give it at checkout, and receives from us your team’s WikiFix id and what you’re buying, with no page content.

WikiFix for Confluence is billed through the Atlassian Marketplace (“Paid via Atlassian”). Atlassian and Notion are where your content lives, not sub-processors: WikiFix reads it from them – through the Forge platform in WikiFix for Confluence, and through the connections your admins set up in WikiFix at app.wikifix.ai – and what we hand Atlassian on our own account is a contact-form submission, as listed above, plus what the WikiFix agent in Atlassian’s Rovo passes to Atlassian’s Rovo AI when someone in WikiFix for Confluence uses it: finding text, page titles and the passages behind them, the text of any page the agent reads for them and, when they settle a finding, the before and after text. Your page bodies are already at Atlassian. It applies on every Data handling level, Strict included, and the agent keeps one summary per person in Atlassian’s Forge storage until they settle the finding or prepare another, with no expiry from us while WikiFix is installed (Atlassian deletes app storage after an uninstall, on its own retention schedule). We don’t sell your data to anyone, ever.

Transfers outside the EU

With EU residency required, no sub-processor on this list receives your page content outside the EU, unless you add your own Anthropic key. Without it, one does: TypeSafe, in the United States, on every Data handling level. If your own review needs the transfer basis, TypeSafe’s Data Processing Addendum incorporates the EU Standard Contractual Clauses (Commission Decision 2021/914) — Module 2 for controller-to-processor transfers and Module 3 for processor-to-sub-processor transfers, which is the one that covers our use of them — together with the UK International Data Transfer Addendum (B1.0).

Two things worth knowing before you rely on that. Their addendum does not contain a clause about training AI models on customer data; what restricts it is the purpose limitation, which permits processing only to provide the service under documented instructions. And their terms set no fixed retention period, so we use TypeSafe with zero data retention: it doesn’t keep the excerpts it receives or use them for anything else, including training AI models. We link their terms rather than summarise them because they are TypeSafe’s to change, not ours.

Sentry and PostHog store their data in the EU, but both are US companies and can process some of it in the US, under the EU-US Data Privacy Framework and the EU Standard Contractual Clauses. We send neither of them page content. Feedback a person writes reaches PostHog as they wrote it.

If none of that is acceptable for your install, require European Union residency and these checks run instead on different AI models, which reside in the EU: eu/jev, operated by Bevelites GmbH on its own server in Germany. Every other AI step stays on AWS Bedrock in the EU, as Data residency explains.

Changes to this list

We publish every addition or replacement here at least 10 business days before it takes effect, so you have time to object before a new sub-processor begins processing your data. You have 30 days from a posting to object on reasonable data-protection grounds — email security@wikifix.ai. If you do, we’ll work with you in good faith to resolve it; if we can’t, you can stop using the Service at any time, and we delete or return your data as the DPA provides.

Follow our sub-processor feed to be notified. Every change is published to it at the same moment it appears here, so you don’t have to watch this page. Entries describe what we intend to do and when.

  • OpenAI added to the list as a model licensor
    Published 4 November 2026 · effective 4 November 2026
    We use an OpenAI model on AWS Bedrock in the EU. OpenAI receives no page content. It is listed for transparency, as Anthropic and Cohere are.
  • Neo4j added to the list: it was missing
    Published 2 November 2026 · effective 2 November 2026
    Neo4j Aura already runs during every scan and should have been on this list. It runs part of a scan's analysis, in AWS Frankfurt, using a numeric index of the content in scope with our internal ids and no page text. A session exists only for its scan run, 24 hours at most if cleanup fails. Nothing about how your data is processed changes today. You have 30 days from this entry to object on data-protection grounds – email security@wikifix.ai.

Our Data Processing Agreement applies automatically when you accept our Terms, so there’s nothing to sign. If your procurement process needs a countersigned copy, email security@wikifix.ai.