AI Transparency Notice

Struxy uses artificial intelligence.

This notice explains when AI is involved, what information may be sent for AI processing, what Struxy does with AI output, and what you should keep in mind when relying on it.

1. You are interacting with AI

When Struxy interprets a capture, extracts tasks or other structured information, suggests relationships, answers questions about your Struxy context, or produces a summary, suggestion or proposed action, an AI system may be involved.

Struxy is an AI product. AI-generated or AI-assisted outputs should not be mistaken for a human assistant independently checking the facts.

2. What Struxy sends for AI processing

Depending on the feature, Struxy may send the AI system:

• the capture, question or instruction you provided;
• relevant source excerpts;
• a limited subset of existing Struxy context selected to help interpret your request;
• system instructions describing the required task and output format.

Struxy is designed to prefer relevant context rather than sending your entire personal workspace to the model by default.

3. Current AI infrastructure

Our current primary inference provider is TensorX Ltd, an Irish AI-inference provider.

TensorX states that its inference service processes prompts and completions using zero-data-retention infrastructure, does not persist prompt or completion content, does not use inference content to train or improve models, and runs inference infrastructure in the European Union.

Struxy may change model or infrastructure providers in the future. Material privacy-impacting changes will be reflected in our Privacy Policy or Third-Party Services page.

4. Struxy may still store your information

TensorX’s zero-retention approach applies to the inference provider’s handling of prompts and completions. It does not mean that Struxy itself forgets the information you intentionally asked Struxy to remember.

Struxy may store your original capture, structured items created from it, reviewable proposals, source evidence, approved relationships, and limited inference/audit metadata.

See the Privacy Policy for retention and deletion information.

5. AI proposals can be wrong

AI can misunderstand what you meant, associate information with the wrong person, infer a date incorrectly, miss an important detail, create an unsupported connection, or produce stale or inaccurate general information.

Do not assume that an AI-generated statement is correct simply because it sounds confident.

6. Source, inference and approval

Struxy’s product design aims to distinguish among:

• source-grounded information — information supported directly by what you provided;
• inferred information — an AI interpretation or suggested relationship;
• general knowledge — information generated from the model’s broader knowledge and which may be stale or incorrect;
• user-approved information — information you have explicitly confirmed where review is required.

The exact interface may evolve during beta. AI output should not silently replace the original source capture.

7. Human review and control

Where an AI proposal could materially change structured personal context, Struxy may require or offer review before treating that proposal as approved information.

Some low-risk organisation or formatting may be automated. Struxy may use confidence thresholds, deterministic checks and other safeguards, but these controls do not eliminate all errors.

You should review anything that materially affects your plans, commitments or decisions.

8. No professional or emergency advice

Struxy is not an emergency service.

Unless we expressly introduce a separately regulated product feature, AI output from Struxy is not a substitute for professional medical, legal, financial, tax, mental-health, or safety-critical advice or judgement.

Do not use Struxy as the sole basis for high-impact decisions.

9. AI-generated content disclosure

Where applicable, Struxy will identify AI-generated or AI-assisted content in a manner appropriate to the product experience.

We also maintain technical and product measures intended to meet applicable transparency requirements for AI systems.

If Struxy later generates media, public-interest publications or content intended for external distribution, additional labelling or machine-readable marking may be applied where required.

10. Questions or corrections

If Struxy produces a wrong or harmful inference, use available correction/review tools or contact support@struxy.app.

For privacy questions: privacy@struxy.app