Skip to content

Concepts

The terms below are used throughout the app, the API and these docs, in the order you meet them on the journey.

TermMeaning
InstanceOne FACE deployment for one organisation, managed by TIDE. After sign-in you choose the instance to work in; the list shows only the instances you hold a grant on.
ConnectionA configured link to one data source: its type (for example Snowflake, SAP, MQTT or a camera) and its settings. Credentials are stored separately and are write-only. See Data sources.
RunnerThe process that reads a data source. The managed runner is part of the server. Self-hosted runners run in your network and enrol with the instance.
TestA connection check. It opens the source with its stored credential, read-only and time-limited, and reads nothing. A connector with no reachability check says so.
Source backingWhether anything FACE ships stands behind a connection. LIVE means nothing of FACE’s stands in, so the connection reads whatever it addresses and a failure surfaces as a failure. Unspecified means the server did not compute it.
RunOne question asked of your data, on Ask or by a brief: FACE reads the sources, answers, then runs the analysis and the twins’ decisions in the background. The app calls it a fetch.
Selected runThe one run that Ask, Options, Why and Actionable Twins show. Pages never borrow figures from another run.
PersonaWho a run or a brief is for: a name, the question its brief asks, the money figure it reads first, and its plays. The product ships five; an instance can reword them or add its own. With none picked, a run uses the Generalist view.
PlayWhat FACE recovers or protects money on: claims recovery, disruptions, routes, fulfilment, cold chain, maintenance, inventory and fleet.
BriefA question FACE asks on its own, at a set time or when data changes, for one persona, telling the people you choose what needs a decision. See Briefs.
OptionOne way to act on a decision, scored on one shared measure with a range. FACE recommends an option only when it is a clear winner. See Options.
CheckA test FACE runs on the rows a run read, such as “Deliveries arrive by their promise”. It holds, fails, or is not measured. See Why.
RuleA business rule FACE found in your systems or you added, checked against what your systems do and against your policy documents.
DecisionA drafted action (a claim, a dispute, an order, a route re-plan) waiting on Actionable Twins until a named person approves or rejects it. Lanes sort decisions into Waiting for you, In flight and Back.
Evidence documentA structured document FACE derives from your data, such as an OS&D report, a temperature-excursion record or a custody audit. The set of document kinds is fixed, and each is built only from the records it cites.
Decision artifactThe full record of one decision, with the drafts it would produce.
Follow-upA flow that calls or messages the people responsible for an item, on rules you approve. It runs by itself only once armed.
RemediationThe work after a compliance, finance or operations finding: assign it, decide it and route it.
TraceThe recorded history of a run: its steps, the sources it read, and what it produced.
Data lineageAn append-only record of which source was read, when, with what shape and outcome. It records structure only, never the values returned.
GroundingRetrieving your own records (Direct Corpus Interaction: lexical search over your own corpus, no embeddings) to support a model’s reasoning, so answers cite your data.