kurozu

Build health context for AI.

Keep a personal timeline of what actually happened: what you tried, what changed, what the result was, and what else was going on in your life.

Download on theApp Store

Local-first. No account needed to start.

Canvas screen/assets/screenshots/canvas.png
1170 × 2532
Canvas. Place facts in time on a vertical timeline.
Fact editor screen/assets/screenshots/fact-editor.png
1170 × 2532
Fact editor. Set the wording, time, and certainty.
Timeline and background screen/assets/screenshots/timeline-profile.png
1170 × 2532
Timeline. Read your facts as a timeline or a plain list.
Export screen/assets/screenshots/export.png
1170 × 2532
Export. Carry your context to any AI.

Design notes

The missing layer

AI systems are getting better at text, health, biology, research, and reasoning. The bottleneck is increasingly not only model capability.

  1. Existing health signals are useful but incomplete/noisy

    Labs, wearables, genetics, clinical records, and trackers all matter. But they are noisy, partial, expensive, burdensome, or context-poor.

  2. Natural-language facts are a good health data format

    Natural text is the native format for nuance, uncertainty, and variable granularity. LLMs make this format newly practical because they are good at reading, grouping, summarizing, transforming, and routing text.

  3. The core object: a good-enough personal timeline

    The app's core is not an AI chat. It is a user-owned timeline and fact layer.

  4. Facts are not interpretations

    This is how natural text stays rigorous: not by pretending all text is objective, but by preserving the line between what was stated and what is suspected.

  5. Extraction AI is thin and optional

    It helps turn messy text or voice into editable candidate facts. It is not the source of truth. The timeline and facts are the source of truth.

  6. The main AI vision is the Experience layer

    The more ambitious AI belongs above the personalization layer. The app should not lock the user's health context into one AI / agent.

  7. Portability

    The Context layer should be easy to export and merge with other signals.

  8. Privacy

    Local-first by default. No login required for the core manual timeline. Export available.

Read the longer writeup →

Roadmap

What's here, and what's coming.

Now

  • Manual timeline and fact editor
  • Point events and approximate spans
  • Local-first storage, no login
  • Export your context as a prompt

Next

  • AI-assisted facts from text and voice
  • Optional cloud sync
  • Import wearable and lab summaries

Later

  • Personal research and doctor prep
  • Bring-your-own-AI agents
  • Open source

FAQ

Common questions.

Local or cloud?

Local-first by default. The manual timeline works fully offline with no login. Cloud sync and AI features are opt-in.

How will pricing work?

The core timeline stays free. Cloud sync and AI features will be paid. The exact details aren't set yet.

Are there AI features in the app?

Yes, but optional. Extraction AI helps turn notes into editable facts. It's never the source of truth, and it only runs in cloud mode.

Will it be open source?

It's on the roadmap, but not the first priority, unless enough people ask for it.

Feedback

Tell us what to build next.

Bugs, ideas, or what's missing for your context. We read everything.