Taskinger1. Architecture Principle: Bring Your Own Key (BYOK)
1.1. Taskinger operates under a strict Bring Your Own Key (BYOK) model. All AI capabilities are completely disabled by default.
1.2. The platform transmits zero user data to AI providers unless you explicitly:
- Input your personal API key from a supported commercial provider (Anthropic Claude, OpenAI ChatGPT, or Google Gemini) in your account settings; and
- Click the "Analyze Note" action button for a specific note or task list.
2. Key Security & Private Storage
2.1. Your API key is stored in an isolated personal Firestore document governed by strict database security rules. It is readable and writable exclusively by your authenticated user ID.
2.2. Workbook members, team colleagues, and system administrators have zero visibility or access to your stored API credentials.
2.3. You may delete your API key at any time in settings with immediate, permanent removal from database storage.
3. Zero-Training Data Policy
3.1. Requests are transmitted directly to the commercial API interfaces of your selected provider (Anthropic PBC, OpenAI LLC, Google LLC).
3.2. Under the standard enterprise/commercial API terms of these providers:
- Data transmitted via API is NOT used to train, retrain, or improve frontier foundation models (LLMs).
- Your task notes are not published or shared with third-party developers.
- Official provider terms: Anthropic Privacy, OpenAI Enterprise Privacy, Google Cloud Privacy.
4. Human-in-the-Loop & No Automated Decision Making
4.1. In compliance with Art. 14 of the EU AI Act and Art. 22 of the GDPR, Taskinger enforces mandatory human oversight (Human-in-the-loop).
4.2. The platform does not engage in automated profiling or make autonomous binding employment, legal, or administrative decisions.
4.3. AI output (such as proposed sub-tasks or note summaries) is always presented as an editable draft proposal. You retain 100% discretion to accept, modify, or discard suggestions before anything is saved to the shared workbook.
5. Content Accuracy & Verification Responsibility
5.1. Large language models may generate incomplete, inaccurate, or hallucinated statements.
5.2. You remain solely responsible for reviewing and verifying all AI-generated suggestions prior to relying on them for clinical, legal, financial, or operational workflows.
6. Contact
For questions regarding AI governance and data privacy:
Email: valentin.k.kirilov@gmail.com
Operator: Vivan Health Projects Ltd., Sveta Troitsa Bl. 376, Entr. G, Apt. 76, 1309 Sofia, Bulgaria.