AI Disclosure
Last updated: May 30, 2026
Nordik Lab uses AI in two places: Nordik Intelligence (the training assistant), and internal processing pipelines such as sport detection and data extraction. This page explains how AI is used, what data is involved, and what limitations apply.
What AI is used
Nordik Intelligence uses Google Gemini (currently gemini-2.5-flash) via the Google Gemini API. Google is a sub-processor for these requests. Data sent to Gemini is subject to Google's API Terms of Service and data protection commitments.
What data is sent to the AI
When you send a message to Nordik Intelligence, the platform constructs a prompt that may include:
- Your message or question
- Summaries of your recent training and recovery data pulled from your connected accounts
- Your saved athlete preferences (training basis, zone models, remembered context)
- Context files you have uploaded to the assistant
- The current conversation history within your session
We do not send your raw GPS tracks, full .fit file contents, or credential data to the AI.
Strava data and AI
The Strava API Agreement restricts the use of Strava API data in AI models. Nordik Lab does not use Strava-sourced activity data to train or fine-tune any AI model.
AI assistant connectors (Claude, ChatGPT)
Separately from Nordik Intelligence (Gemini, above), Nordik Lab supports connecting your account to Claude (Anthropic) or ChatGPT (OpenAI) via the Model Context Protocol (MCP), from Settings → API Keys. This is opt-in and off until you connect it yourself. Once connected, that assistant can call back into Nordik Lab to read — and, depending on the scopes you grant, write — your training data as part of your conversation with it. Anthropic or OpenAI processes that exchange under their own API terms; Nordik Lab does not control or see how Claude or ChatGPT itself retains or uses that data beyond the API call.
Coach connections: coaches can connect their own Claude/ChatGPT assistant to their coaching account. That assistant can only see or add notes to a given athlete's data if the athlete has an active coaching relationship and has separately turned on "Coach AI Access" for that coach in their own Settings. This defaults to off, is checked on every request (not just once at connection time), and takes effect immediately if turned off. Athletes can see this setting, and a log of which of their data a coach's connected assistant has accessed, from Settings → API Keys.
No training on your data
Nordik Lab does not train, fine-tune, or otherwise use your personal data, health data, or training records to improve or build any AI model. Data sent to the Gemini API is processed under Google's standard commercial API terms, which include commitments that API inputs are not used to train Google's general models without your consent.
Memory and preferences
Nordik Intelligence can remember athlete preferences and context across conversations. Remembered preferences are stored in your athlete profile and can be reviewed, edited, pinned, or deleted from Settings → Preferences at any time. Uploaded context files can also be managed from that page.
Limitations and responsibility
AI outputs may be incomplete, incorrect, or unsuitable for your situation. Nordik Intelligence is designed to produce editable drafts and ask clarifying questions before creating materially underspecified artifacts — but it can still be wrong. You must review all AI-generated workouts, training blocks, charts, and recommendations before acting on them.
Nordik Intelligence is not a medical device and its outputs do not constitute medical advice. Do not use it to make emergency health decisions or as a substitute for professional sports medicine or clinical care.
EU AI Act
Nordik Intelligence is a general-purpose AI assistant used for athletic training management. It does not make decisions with significant legal or similar effects on individuals, does not assess creditworthiness, and does not operate in any high-risk domain as defined under the EU AI Act. It is considered a minimal-risk AI system.
