Your AI, integrating health, home, family, community.
Is your home making you sick?
Where you live, how you live, and the people around you collectively determine much of how long you stay healthy. This assessment estimates how much — by category, with a confidence interval on every number, and with the source for each claim.
What this is, and what it is not
It is a population-attributable risk estimate: it reads your circumstances against population baselines and the published evidence, then reports where your years are being added or subtracted. It is not a prediction about you personally, and it is not a date. Anyone offering you a single number without a range is selling certainty that does not exist.
It requires no blood, no saliva, and no wearable. Those measure your body; this reads the environment your body lives in — which is the part almost nobody measures, and the part you can actually change.
On the research this draws from: the demography behind the best-known longevity regions has been seriously challenged, and we wrote up what survived that audit and what did not rather than leaving it unsaid.
This assessment is for one person. If you run a residential community and want the same measurement applied to the whole place — see whether your community is a match.
What you get
You answer a structured set of questions about your life — most of them optional, all of them yours. We turn your answers into a personalized assessment that breaks your life expectancy down by category and tells you exactly which actions move it.
A baseline number
Based on your country, sex, and current age, we start from a population baseline (sourced from WHO life-expectancy tables).
Six attribution categories
We break the difference between baseline and your personal estimate into lifestyle, home environment, community + social fabric, family, work + purpose, and health history. Each shows years gained or lost with confidence intervals.
A ranked action plan
Five to seven concrete actions ordered by impact-per-effort, each with an estimated year-impact, difficulty, and time horizon to first effect.
Provenance you can read
When a claim is backed by an evidence-validated entry in our doctrine graph, we show you the source. When it's a frontier-model estimate based on general knowledge, we say so.
Why this works.
The fair questions to ask before trusting any system with this kind of data.
General-purpose models reason in the moment from whatever you paste. Ikaria reasons against a calibrated substrate — a graph of validated longevity attributions, signed effect sizes with confidence intervals, and provenance per claim (which paper, which population, what its limitations are). When the substrate doesn't have a hit, we say so explicitly with a "frontier-estimate" badge, instead of dressing up a plausible guess as authority. You get reasoning plus bookkeeping — not just reasoning.
A few honest things up front
- This is not medical advice. Confidence intervals are wide because life-expectancy attribution from a self-reported intake is genuinely uncertain.
- Your inputs are stored. We use them to build your assessment and to improve the substrate over time. They're not sold, traded, or used for advertising — see privacy when we publish it.
- Most users finish in 8-12 minutes. Skip any field you don't want to answer; the assessment widens its uncertainty bands automatically.