Cookies, briefly

Focarly uses essential cookies to keep you signed in. We don't use marketing trackers. Analytics help us make the product better — your call. Details.

AI Disclosure

How we use AI in Focarly, your consent, and AI's limits. Last updated: April 2026.

1. What is AI in Focarly?

Focarly is an AI-powered tool. When you interact with the coaching feature, you are communicating with an artificial intelligence system (specifically, Anthropic's Claude models: Opus 4.7, Sonnet 4.6, and Haiku 4.5), not a human.

2. What AI can and cannot do

AI in Focarly CAN:

  • Generate helpful coaching-style responses
  • Help you break down tasks
  • Remember context within a conversation
  • Point you toward useful resources

AI in Focarly CANNOT:

  • Diagnose medical conditions
  • Prescribe medication or treatment
  • Replace therapy, coaching, or medical care
  • Always be accurate — AI can “hallucinate” or make factually incorrect statements
  • Provide human judgment or empathy

3. Your consent to AI processing

By using Focarly's coaching feature, you consent to:

  • Your messages being processed by AI models from Anthropic PBC
  • AI-generated responses being displayed to you
  • Your conversations being logged for safety and service improvement

Withdrawing consent. Under GDPR Art. 7(3), withdrawing consent must be as easy as granting it and cannot be conditioned on continued payment. To withdraw AI-processing consent without canceling your subscription:

  • Email privacy@focarly.comwith the subject line “Withdraw AI consent”. We'll disable AI processing on your account within 7 days and pause billing at the end of your current cycle with your permission (the product becomes read-only without AI — you'll still have access to your history and settings).
  • Alternatively, request full account deletion from Privacy & data settings. Your subscription cancels and all data hard-deletes within 30 days (after the 7-day grace period).

4. How AI is used

  • Coaching conversations: Claude Sonnet 4.6 generates personalized coaching responses based on your messages.
  • Safety filters: Claude Haiku 4.5 analyzes every message and response for crisis situations, medical advice requests, or inappropriate content.
  • Weekly insights: Claude Opus 4.7 analyzes your weekly usage patterns to generate (non-diagnostic) pattern insights.

5. AI limitations and appropriate use

AI outputs may be inaccurate, incomplete, out-of-date, reflect biases from training data, misinterpret your messages, or suggest actions not right for your situation.

Use your own judgment. Do NOT rely on AI as the sole basis for important decisions, especially medical, legal, financial, or safety-critical decisions.

6. No human impersonation

We always clearly identify AI-generated content. Focarly will never pretend to be a human therapist, doctor, or coach.

7. EU AI Act compliance

Focarly is classified as a “limited risk” AI system under the EU AI Act. We comply with Article 50 transparency obligations: users are informed they are interacting with AI.

We do NOT deploy:

  • High-risk AI systems (e.g., for employment, healthcare diagnosis)
  • Prohibited AI practices (e.g., emotion recognition, social scoring)

8. Your choices

  • Stop using the AI coaching feature at any time
  • Delete all AI interaction history from your account
  • Opt out of optional features involving AI
  • Contact us with concerns about AI outputs: safety@focarly.com

9. Voice features

If you turn on voice in chat, one additional AI processor is involved:

  • ElevenLabs Inc.(US) — handles voice both ways: transcribes the voice messages you send (Scribe) and synthesizes the coach's spoken replies from text. May retain transmitted audio and text per their published Data Processing Addendum.

Focarly does not store audio on our servers. Before voice is activated, the chat toggle shows the specific vendor disclosure and you explicitly consent — that consent record (with version) is logged so we can show exactly what you agreed to.

Voice input features are temporarily unavailable in Illinois pending a finalized BIPA-compliant consent flow. Voice output (replies) is available everywhere — and text-only chat works for everyone, in every region.

You can disable voice at any time via the toggle in the chat interface. Disabling it revokes consent for future voice processing; it does not affect your subscription or text chat. For the full data-handling details, see Privacy Policy → AI processing.

10. Contact

AI safety concerns: safety@focarly.com
General AI questions: hello@focarly.com