AI Processing Notice

What AI does in Legacy Pulse

Legacy Pulse uses AI for two narrow jobs. First, when you ask it to organize an uploaded document, AI reads that document and extracts structured entries — lab values, medications, immunizations, vitals — each linked back to the exact source document it came from. Second, when you ask a question, AI searches your family's own records and composes an answer from them, with citations to the source documents so you can check every statement.

The AI works only from records you uploaded. It does not browse the internet for answers about your health, and it does not mix your records with anyone else's.

What the AI will not do

The AI in Legacy Pulse does not diagnose conditions, recommend or adjust treatment, suggest medication doses, or assess how urgent a situation is. If you ask it a question that calls for medical judgment, it will decline calmly and suggest discussing the records with a clinician. If you think someone may need urgent care, contact a clinician or emergency services — Legacy Pulse cannot assess emergencies.

The AI describes what your records say, using observational language: a value "was recorded" or is "outside its reference range." It does not tell you what a value means for your health.

Where the processing happens

AI processing runs server-side on contracted infrastructure. When you trigger it — by organizing a document or asking a question — the relevant content is sent to our AI provider, processed to produce the entries or the answer, and returned to your vault.

Our vendor terms prohibit the provider from using your data to train AI models and limit how long the provider may retain it. Your records are never used to train AI models — not ours, and not the vendor's. Processing happens only when you ask for it: we do not run AI over your vault in the background or speculatively.

Confidence, mistakes, and your right to correct

AI extraction is good but not perfect. Every extracted value carries a confidence level (High, Medium, or Low), and values the AI was less sure about are flagged for optional review. The original document is always preserved exactly as you uploaded it, so you can compare any entry against its source with one click.

If an extracted value is wrong, you can correct it or mark it as disputed. The dispute or correction stays attached to the value everywhere it appears — in search results and in summaries — so an AI mistake never quietly becomes the record.

Your consent, and how to withdraw it

AI processing happens only with your consent, which you are asked for when you first use an AI feature. You can withdraw that consent at any time in Settings.

When you withdraw: organizing new documents with AI and asking questions both pause immediately. Nothing is removed — your original files stay safely in the vault, and entries that were already extracted remain (you can delete any of them, or their source documents, whenever you choose). If you grant consent again later, the paused features simply resume.

Withdrawal is recorded with a timestamp, just like the original consent, so there is always a clear record of what you agreed to and when.

Contact

Questions about this notice, or requests you would rather not make in the app, can be sent to privacy@lgc.my. You can also exercise most of your choices directly in Settings.

We respond to privacy requests within the timelines required by applicable law, and we will always tell you what we did in response.