Data residency
Storage and compute run in AWS Frankfurt (eu-central-1). AI steps run on AWS Bedrock through EU inference profiles, which can serve a request from any EU region, apart from two analysis checks. Without EU residency, those two run on TypeSafe in the United States, on every Data handling level. A request to TypeSafe receives excerpts of the pages it checks, which can come 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; it never receives the whole space. 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.
Need those checks in the EU too? Require European Union residency during onboarding, or later on the Residency tab – in WikiFix Admin in WikiFix for Confluence, or under Admin at app.wikifix.ai. Those two checks then run 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, so every AI step is in the EU. eu/jev receives the same excerpts TypeSafe would, doesn’t store them, in its database or its logs, and doesn’t use them to train AI models. Scans use the same credits either way. We administer WikiFix from Moldova, as the DPA explains.
The options
Section titled “The options”| Setting | What it means |
|---|---|
| No requirement (default) | We pick the best option for each step, wherever it runs. Today that is the EU for everything but the two US checks. It promises nothing: your content may be processed outside the EU. |
| European Union | Every AI step runs in the EU: the two analysis checks on eu/jev in Germany, the rest on AWS Bedrock in the EU. We administer WikiFix from Moldova, as the DPA explains. |
| United States | We cannot serve this today. Our infrastructure is pinned to Frankfurt, so a US requirement is refused rather than served from the EU and labelled US. You can record it; scans stop until we can meet it. If that is a deal-driver, tell us at security@wikifix.ai. |
A requirement we cannot meet stops the scan
Section titled “A requirement we cannot meet stops the scan”The scan stops before any page content is read or sent to a model, and nothing is billed; we only confirm the connection and refresh the list of spaces. It does not fall back to a location elsewhere and carry on. The run is marked failed with the reason, and the Residency tab shows it.
The other route: your own key
Section titled “The other route: your own key”On a paid Advanced plan, bring your own Anthropic API key: the AI work runs on Claude under your Anthropic account and spends no credits, whatever residency you’ve set. That includes the two analysis checks, so while a scan runs on your key nothing goes to TypeSafe or eu/jev; with credits first, the usual path applies until the credits run out. Indexing stays on AWS Bedrock, and those calls are still recorded to Langfuse as your Data handling level sets. That is outside what this setting controls — Anthropic processes it under your agreement with them, wherever they do. See Credits & capacity.
Storage is always AWS Frankfurt, whatever you set. Residency governs where your content is processed, not where it is kept.
Related: Security · Security & data · Sub-processors · Data Processing Agreement