Choose your role
Select the stakeholder mode that matches how you will use the briefing.
FYNURA DOCS
Use Fynura to discover, verify, understand, monitor, and act on authoritative public-health evidence.
Fynura is evidence-grounded public-health intelligence. It retrieves structured observations from approved authorities, preserves provenance, and clearly labels uncertainty. It is not a diagnostic tool or a substitute for official authorities.
Select the stakeholder mode that matches how you will use the briefing.
Choose measles, Ebola, or cholera and inspect the reporting cutoff.
Ask what changed, what is known, or which source supports a claim.
Open the evidence drawer and follow observation links to the authority.
Use charts and maps with their units, scope, and limitations.
Compare dated observations; do not interpret missing data as zero.
Create a watch for the configured geography and threat.
Use exported visuals with their citations and reporting dates intact.
The global situation shows evidence coverage. Active threat cards summarize the latest verified observation. Ask Fynura answers only from the structured evidence store. Evidence drawers reconstruct each claim.
Map markers represent verified reported activity or configured coverage. Bubble color identifies the health threat and size represents reported case magnitude, not population-adjusted incidence.
Ask concise questions such as “What is the latest Ebola situation?” or “Which source supports this case count?” Unsupported clinical or personal-risk questions are declined.
Always read the title, unit, reporting cutoff, geography, and limitations. A map showing no marker does not establish zero activity.
Approved sources include WHO and configured public-health authorities. A configured source is not counted as corroboration until its observation has been retrieved and verified.
Choose Watch on a threat card. The MVP records the watch in the running service; durable notification delivery requires Firestore, Pub/Sub, and Scheduler configuration.
Google ADK coordinates discovery, extraction, verification, epidemiology, and communication roles. Gemini runs through Vertex AI. Numeric extraction, arithmetic, and evidence selection retain deterministic safeguards.
Evidence confidence summarizes source authority, recency, completeness, and agreement. HIGH does not mean certainty; it means the stored claim has strong support under the current rules.
Fynura distinguishes event, reporting, publication, and retrieval dates. Surveillance may be delayed or revised after publication.
Fynura does not silently average conflicting values or sum overlapping reports. Conflicts remain visible, and the canonical view favors authoritative, scope-compatible observations.
Fynura does not replace public-health authorities, clinical judgment, diagnosis, or medical advice. Surveillance data may be provisional, incomplete, delayed, or revised.
Source availability varies. Automated access can be restricted. Current coverage is not exhaustive, and the MVP may use in-memory state that resets on a new Cloud Run instance.
The proportion of reported deaths among reported cases for the same compatible scope and period. It is not an individual prognosis and can be biased by reporting delays.
No. It is near-real-time relative to accessible authoritative reporting.
No. Missing data is never treated as zero activity.
Build 0.3 · 29 Aug 2026 · Added Fynura Docs, searchable guidance, privacy notice, lightweight onboarding architecture, and privacy-minimized product events.