Prepare the case
Bring different sources into a form PI can use and organize them into a structured starting point.
What PI Is
PI helps make sense of one changing case.
PI is built on causal AI and designed for within-person reasoning. It tries to understand what may be shaping one person's current picture over time, not just what seems to show up next to it.
The core idea
PI looks at one person's story over time, not a population average.
Instead of reducing a case to one score or borrowing the main answer from broad population patterns, PI tries to keep the order, context, baseline, and cause-and-effect clues that matter within one person's history.
What PI can tell you
- What may be causing the current result
- How the current picture compares with that person's usual baseline
- What may be helping versus hurting
- Which factors may be working together
- Which less obvious factors may matter more than they seem
- What may matter most right now
- What looks like a weak point or central pressure point
- Which action, if any, deserves consideration first
- Where PI thinks caution is the right answer
Staged workflow
PI does not jump to a final answer.
PI moves through stages inside one case. Stage 1 forms early hypotheses. Stage 2 refines candidate structure. Stage 3 confirms what holds up. Claims only move forward when there is enough support, which helps PI stay useful with patchy data and honest about what is still provisional.
Stage 1: form early hypotheses
Make a first pass when the case is new, the data is limited, or rediscovery is needed.
Stage 2: refine candidate structure
Strengthen, reshape, or drop candidate links as more support appears.
Stage 3: confirm what holds up
Confirm, downgrade, or keep findings provisional depending on the evidence.
Report and select with boundaries
Show what PI sees, what is confirmed, what is still tentative, what matters most, and whether PI should stop at explanation or surface a permitted action.
Principles
PI is designed to stay honest.
Within-person first
PI is built to reason within one person's case over time. Cohort evidence can help evaluate PI, but it is not the primary reasoning target.
Cause, not coincidence
PI tries to understand what may be causing the result, not just what seems to show up beside it.
Different stages, different trust
Early hypotheses, refined candidates, and confirmed findings do not get treated as if they mean the same thing.
Works with patchy data
PI is designed to stay useful when the case is incomplete, without pretending the missing pieces are known.
Action is selected, not guessed
PI does not turn every plausible lever into advice. It should surface a permitted action only when the governed lane clears.
FAQ
Straight answers.
Is PI a diagnostic engine?
No. PI is currently being positioned as a careful support layer, not as an autonomous diagnosis tool.
Does PI jump straight to a confident answer?
No. It moves from early hypotheses to candidate structure to confirmation, and it can keep findings provisional when the support is not strong enough.
What does PI actually help you see?
PI helps surface likely causes, helpful versus harmful influences, connected chains, less obvious levers, weak points, what may matter most within one case, and which action lane, if any, deserves consideration.
Is PI population-based?
No. PI is a within-person, N=1 reasoning layer. It centers one person's baseline, change, and current structure rather than starting from population-level averages.
Can PI work with sparse data?
Yes. PI is designed to stay useful with sparse or patchy data, keep weaker findings provisional, and abstain when the support is too thin for a stronger claim.
Where is PI strongest today?
As a health-first, within-person layer for partner, pilot, research, and product settings.
Continue
Explore PI in health.
PI Health is currently the strongest expression of the platform and the right place to understand its near-term fit.