Hypothesis & Threads
Inside a case, threads are where the thinking happens. They come in two kinds: hypotheses — explanations you’re testing — and discussions — free-form notes and debrief. A hypothesis is the heart of it: you state a theory, then let the evidence confirm or kill it.
How a hypothesis evolves
A hypothesis starts as a proposal and moves as the evidence comes in:
| Verdict | What it means |
|---|---|
| Proposed | Stated but not yet tested |
| Supported | The evidence consistently backs it |
| Refuted | The evidence consistently contradicts it |
| Inconclusive | The evidence is mixed or insufficient |
Because you can run several hypotheses side by side, a case becomes a set of competing explanations — and the evidence shows which one is winning.
Linking evidence: for and against
The power of a hypothesis comes from connecting it to evidence with a clear stance:
| Stance | Meaning |
|---|---|
| Supports | This evidence backs the hypothesis |
| Contradicts | This evidence argues against it |
| Neutral | Relevant context, but not pointing either way |
Each hypothesis shows a running for / against count, so at a glance you can see how strongly the evidence leans — and which theory is holding up.
Confidence
Alongside the verdict, you can set a confidence level — your own certainty in the hypothesis. It’s deliberately separate from the verdict: you might have a supported hypothesis you’re only moderately sure of, or a refuted one you’re very confident is wrong. Verdict captures what the evidence says; confidence captures how sure you are.
A record of the reasoning
Every thread keeps a running history — the statements added, the verdict and confidence changes, and the notes along the way. That makes a hypothesis not just a conclusion but a transparent line of reasoning anyone can follow later, including which steps were taken by a teammate and which by Autopilot.
Hypotheses and their evidence links are also drawn on the knowledge graph, so you can see which evidence supports which theory.