Reporting

Objects — Event · Case state · Task — read live over the Record

Reporting is how a deployment sees the operation. Because state and the full event history are one queryable store, reporting reads the operation live — off the same record the work runs on, never a nightly export to a separate warehouse. A deployment runs a standard set of reports and dashboards and lets a privileged user query the store directly for an ad-hoc cut; both read current state, so where every case is, who holds it, and how long it has sat is answerable for the whole operation at once, in real time.

Live, off the record

Every queue movement, status change, assignment, actor, and dwell time is already an append-only event on the case across its whole lifetime. Reporting reads that store directly, so the figures a governing statute or reporting regime asks for are captured as the work happens rather than reconstructed after it. There is nothing to reconcile between a “real” store and a reporting copy — audit, retention, and analytics all read the same case-keyed events.

The standard reports

The platform-provided baseline is a set of reports and dashboards rendered in the application and queryable as the operation runs. They cover what every adjudication operation watches:

ReportWhat it showsRead by
ThroughputCases opened vs. closed per period, broken down by case type and laneOperation lead, supervisor
Age in each stepHow long cases sit at each workflow step — median, 75th, 95th percentileOperation lead, supervisor
Overdue casesCases past their workflow deadline, grouped by assignee and stepSupervisor, operator
Queue depth and agingOpen task count per queue and the oldest item’s age in eachSupervisor
Reviewer loadOpen tasks per assignee, broken down by status (assigned, in progress, on hold)Supervisor
Distribution outcomesDistribution runs over time — what got assigned, by lane and assignee, with the lever set on eachOperator, operation lead
Decisions per issue typeDisposition breakdown by issue type and categoryOperation lead, quality reviewer
Session dispositionsHeld / no-show / withdrawn / postponed rates by session typeOperation lead
Work-product evaluationsAuthor scores over time along the deployment’s defined factorsQuality reviewer, supervisor
AI disposition ratesPer-model accept / edit / reject rates for each suggestion kind (Issue, Task, relevance)Operation lead, operator

Quality and oversight

Quality review is a task that scores an author’s work product along defined factors; those scores read back here, so complexity and quality per reviewer sit on the same record as the work — not a spreadsheet on the side. An operator reads the operation itself: the configuration changes recorded over a period, the distribution runs that completed or errored, and the integration adapters reporting healthy or failing.

Ad-hoc, over live state

Beyond the standard set, a privileged user queries the store directly for a cut the standard reports don’t carry — a slice by party, by issue type, by adapter, by any data element the operation categorizes by. Because the query runs over live state, the answer is current as of the moment it’s asked.

Integrate or provide

The reporting tool itself is integrate-or-provide. A deployment binds the reporting port to a BI tool it already runs — the port serves reads (current state and the event stream) on a pull schedule the tool sets, and exposes the data elements the operation categorizes by so the BI tool’s dimensions match what authors and operators see — or it uses the platform’s own reports and dashboards. Either way the source is the one live record. Role-tailored dashboards put the high-level numbers in front of the people who run the operation, gated by the role each holds.