About contextualisation
Contextualisation is how Tiro Pro works out what your documents mean. It is built around matters: collections you have declared to mean "these documents belong to one evolving story". File documents into a matter and an AI pass reads them as a group — what each document is, how they relate to one another, and what the set amounts to — then keeps that understanding current as the matter changes. You never run it by hand: adding a document to a matter is the whole instruction. To set one up, see Contextualise documents.
Matters and ordinary collections
An ordinary collection is just a grouping — it says nothing about whether its documents are related, and grouping them changes nothing. A matter is different by declaration: its members belong together, and the system keeps their context current. You create one by ticking This collection is a matter when creating a collection, or by promoting an existing collection later. Everything else about collections still applies: a matter can hold sub-collections for filing (their documents count as part of the matter), a document can belong to several matters at once, and matters sit inside ordinary collections freely. The one forbidden shape is a matter inside another matter — the app will explain and stop you.
What contextualisation produces
- A role for each document — what kind of document it is: a contract, a note, a data sheet, and so on. Shown in the documents table's Type column.
- A description for each document — a single factual sentence: what the document is, the parties, and what it covers. Shown in the Description column and in the details panel.
- The relationships between the documents — which document amends, supports, or annotates which. These are the links the navigator draws, and they name the matter that drew them.
- A journal entry per review — a short narrative of what that review established: which documents it worked in, how they connect, and anything it revised. The journal is read from the matter itself (see the guide), interleaved with the matter's membership changes — documents added, removed, deleted, or restored — so it reads as the matter's history, not a document in your library.
Contextualisation writes a description for each document it works in, replacing what was there — including one you typed yourself. Documents it has already settled keep their role and description on ordinary runs.
How a matter stays current
Contextualisation reads the set as a whole, and it works incrementally. When documents join a matter, the system waits briefly for the batch to settle (and for each newcomer to finish processing), then runs once: newcomers are woven into the existing picture, and what was already settled is left alone. Occasionally a new document genuinely changes how existing documents relate — an amendment that restates an agreement, for instance — and the run may then revise exactly the settled links that the newcomer affects, never more. The journal records every such revision, and links you have drawn yourself are never touched by any run.
When a document leaves a matter, its machine-drawn links to the rest of that matter are removed at once — unless another matter also relies on them, in which case they stay. Your own links always stay.
A matter can also be made dormant: it keeps everything it has, but stops watching for changes — useful for a concluded matter. Reactivate it and, if documents changed while it slept, one catch-up run brings the context up to date. And if you ever want a fresh opinion of the whole set, Re-contextualise matter documents de novo re-reads everything and converges the picture on the new answer.
Where the context is used
The roles, descriptions, and relationships feed the rest of the app: the documents table and details panel show them, the navigator draws them, each run appears in Activity, and the agent in discussions can look up how your documents relate when it answers.