AI and Your Data

The short version

Atlas uses Anthropic's Claude models for specific analyses. Only content you have entered yourself is sent for AI processing. Anthropic does not retain your data and does not use it to train models. All AI-generated outputs are stored in your EU database.

Which AI features does Spekir use?

AI is an optional layer on top of Atlas. The following features make AI calls:

Strategy parsing — structures uploaded strategy documents
Capability analysis — generates and classifies organisational capabilities
Technology lifecycle — assesses maturity and risk for applications
Risk classification — screens systems against EU AI Act Annex III
Narrative generator — formulates executive summaries of portfolio data

Features such as search, filtering, and report export do not use AI.

What data is sent to the AI?

Only content from your own workspace is included in AI calls — for example application descriptions, technology names, and strategy text that you have entered or uploaded. Data from other workspaces or users is never included in an AI request from your account.

Anthropic's data handling

Spekir uses the Anthropic API under the following terms:

API data is not used to train Anthropic's models
Data is processed transiently — Anthropic does not store the content of API calls
Anthropic's usage policies prohibit sharing customer data across accounts

Source: Anthropic API usage policy and data processing terms (anthropic.com).

Where are AI outputs stored?

AI-generated outputs — such as classifications, narratives, and risk assessments — are stored in your workspace database on Neon in Frankfurt (EU). They are scoped to your workspace and can be exported or deleted like any other Atlas data.

Your controls

AI features are opt-in: your administrator can disable them for the entire workspace
All AI-generated data is deleted if you exercise your right to erasure
You can export all your data — including AI outputs — from account settings