Lab / Researcher · Building
Research the case
without naming it.
Storyline's research desk, becoming its own app for iPhone, iPad and Mac. A desk holds many research projects, each with its own protected names, vault and audit chain. Questions go to the web generalised, and every conclusion shows what it rests on.
Try it: ask without naming
The question still reads. The names never leave.
Each project has a vocabulary of protected names, and each name has a treatment: a person becomes their role, a place is widened, a fact becomes its generic key, and a matter's title is dropped. You see both texts and approve; the screen still checks what you approved before it can go.
| Name | Kind | Leaves as |
|---|
This page runs the generaliser's rules in your browser and sends nothing: longest name first, Unicode-aware boundaries so "Rao" never matches inside "Raoul", possessives kept ("Rao's" becomes "a former employee's"), and no double spaces or stray commas left where a name was dropped. In the app, generalising is a courtesy to the writer; the security boundary stays the egress screen, which checks the approved text again.
Try it: how complete is it, really?
A number you can rebuild from what's on screen.
Every research line is a set of sub-questions: answered counts 1, partial 0.5, open 0. A line's score is its average, and a project's is the average of its lines, so one long line can't drown a short one. Tap a status to change it. Accepting an open thread from the assistant lowers the number, because the line just gained a question it can't answer.
Supersede a source and everything that rests on it lights up: the finding, the inference built on it, and the conclusion built on that. Showing only the direct citation would understate the damage.
How it's built
One desk, many projects, nothing shared by accident.
Desk, project, line
A desk holds projects across fields: a legal matter, a clinical question, an investigation, a due-diligence target. Each line keeps its searches, the documents it read and what each one let you infer.
Screening it already trusts
Detection and blocking come from Storyline's audited egress screen and audit chain. Researcher adds the rewrite that turns a blocked draft into a question that can go.
Numbers derived, never stored
Completeness and fragility are computed from the trail each time, and tested without a database, a store or an unlocked vault.
In the lab: Researcher is not released and nothing here is a shipped claim. How Storyline's research screening works today →