Private evidence for everyday health

Know your why.
Shape what’s next.

Vueniverse turns the health signals and recurring moments already in your life into evidence you can replay, question, and act on, without handing an AI the power to invent the result.

  • Android-first
  • Local-first
  • Evidence before language
Vueniverse constellation logo over a flowing universe of light
Your universe of signals. One evidence trail.
01
Observe clearlySee what the app actually has.
02
Question the patternCounts, limits, and counterevidence stay visible.
03
Test one changeTurn an observation into a personal learning loop.

The evidence loop

From a faint signal to something you can use.

Vueniverse keeps the entire path visible. Select a stage to see how a pattern earns your trust. Then see what happens next.

Start with the source

See your data before it becomes a finding.

Review the 30-day rhythm, signal coverage, source mix, and privacy-safe records. Gaps stay visible; Vueniverse never fills them with examples.

What you can inspect Exact local counts, coverage, and source freshness

Evidence, not a wellness score

A finding that shows its work.

Every visible result starts with deterministic comparison. You can see what repeated, what was excluded, what contradicted it, and how complete the data was.

Vueniverse found
“Your heart rate was usually higher before your recurring 1:1.”
Fictional Snapshot data · not medical advice
Moment fingerprint

Recurring 1:1

Supported
+14+70
Repeated6 of 8comparable meetings
Difference+8–14bpm before start
Excluded4missing or unreliable

Counterevidence stays visible. Two comparable meetings did not show the same rise.

Two lanes. One truth boundary.

The numbers come first. AI gets the last word, not the first.

01 Deterministic evidence

Code calculates the result.

Vueniverse normalizes local records, matches comparison windows, measures effects, and records exclusions and uncertainty before any narration.

  • Matched personal baselines
  • Visible completeness and exclusions
  • Versioned, recomputable evidence
02 Bounded MedGemma

AI explains only what passed.

MedGemma can turn the calculated evidence into plain language on device. An output guard checks every answer; safe reviewed wording takes over if the model is unavailable or invalid.

  • Evidence-grounded explanations
  • No diagnosis or treatment answers
  • Checked fallback always available

Privacy by architecture

Your universe stays yours.

Personal records live in an encrypted, app-private local database. Vueniverse reduces sensitive context before analysis and keeps real and fictional data in physically separate stores.

01
Encrypted local timelinePersonal records stay in app-private storage.
02
Identity-free calendar contextKeep category and timing, not names, titles, attendees, or descriptions.
03
Live and Snapshot isolationExplore safely without mixing fictional and personal data.
04
Deletion has consequencesChanged source data invalidates and recalculates affected findings.

Honest product status

Built to show what is real, and what is still being proven.

Working foundation

Evidence remains useful without the model.

The encrypted timeline, Android sources, deterministic analysis, replay, challenge, experiments, history, and proof export form the reliable core.

Acceptance in progress

Phone-local MedGemma still has physical gates.

Full model delivery, real-phone performance, memory, battery, and thermal acceptance must close before on-device narration is treated as final.

Vueniverse

Know your why.
Shape what’s next.

Explore the project