Trust Overview

Outercite is a platform built on citation data, and that data is only useful if you can trust it. This section explains how we think about data handling, how client workspaces are kept isolated from one another, and how the verification pipeline keeps false positives near zero.

What you'll learn

• The three pillars of Outercite's trust model: data minimisation, workspace isolation, and verified accuracy
• Where to find detail on each pillar
• How to reach us for security-specific questions

The three pillars

Data you can rely on

The verification pipeline runs every AI response through two independent models before counting a citation. Only citations that both models agree on are recorded. This design keeps false positives near zero. When your dashboard says you were cited, the data behind it is genuinely trustworthy.

The confidence score on each citation is a direct output of that consensus step. A higher score means stronger agreement between the two models. A lower score means the citation exists but had some ambiguity in context or prominence.

Across the prompts we track, brands are cited in about half of all checks (a 48% citation rate), and every counted citation passed through the full four-stage pipeline.

Data you control

Outercite stores the prompts you choose to track, the AI responses those prompts generate, and the citation results that come out of the verification pipeline. It does not reach into your brand assets, your CRM, or your website content. You define what is tracked by setting up your prompts in the dashboard.

The data handling and privacy page explains in detail what is collected, what it is used for, and the principle of data minimisation we follow.

Data that stays in your workspace

For agency customers, each client workspace is isolated at the data layer. One client cannot see another client's citation data, keywords, competitors, or reports. The public Outercite knowledge base contains no client data at all.

The data isolation page explains the multi-tenant model in full.

The verification model in brief

The pipeline has four stages: intent classification, deep analysis by one model, independent cross-verification by a second model, and a consensus step. Only the consensus result is counted. The how we verify citations page covers the trust and accuracy angle of that design.

Security questions and documentation

For questions specific to your organisation, including documentation requests, data processing agreements, or sub-processor information, contact the Outercite team directly. We do not publish blanket security claims here. We would rather answer your specific question accurately than make a general assertion that may not apply to your context.

Try this in Outercite

To see the pipeline and your own citation data in one place, open your dashboard. Each citation shows the confidence score and a proof snippet so you can verify the result yourself.

Related