MyAlivea

01What it does

Health apps are good at counting steps and bad at knowing you. MyAlivea holds a profile — labs, metrics, goals, history — and answers against it, so the reply to "should I be worried about this ferritin number" is grounded in your previous results rather than a generic article.

02What's inside

• Lab results and metrics tracked over time, with trend context on each value
• Conversational chat with long-term memory: past conversations, labs and learned preferences carried into every reply
• Tools the chat calls while it answers — lab lookup, conversation search — instead of free recall
• Agent architecture: event-driven reactions plus thirteen scheduled jobs, including a daily health scan
• Pattern detection across signals rather than single-value thresholds
• Preference learning — how much detail, how blunt, which channels
• Interaction tracking so advice does not repeat itself

03How it works

An event-driven backend with a worker running thirteen scheduled jobs alongside the chat service. Health data stays in Postgres on the same host; nothing is sold, shared or used to train a model. Twenty-five unit tests cover the agent core.