Jobs

The Jobs page is the persistent home for every match job you've ever run. It has two views: a list of all jobs and a detail view for any individual job.

Jobs list

A sortable table of every job, with columns:

Label / ID — your label or an auto-generated one
Status — completed, processing, failed, pending
Source file and Master file — what was matched against what
Match rate — percentage matched (with color: green ≥80%, amber 60–79%, red <60%)
Row count — source rows processed
Tags — labels you applied
Date — when the job started

Filtering and sorting

Filter by status, source/master file, tag, or date range. Sort by any column. The filter state is preserved as query params so you can bookmark specific views.

Bulk actions

Select multiple jobs to:

Delete (with confirmation)
Tag with a label
Download an ad-hoc summary across the selection

Job detail

Click any job to open its detail page. This page is the single source of truth for everything the match produced — results, engine telemetry, Genie's narrative, and the review queue.

The job's label, status, timestamps, and processing time. If the job is still running, a live SSE progress bar appears here too.

Summary cards

Same four stat cards as the Review step: source rows, matched, needs review, unmatched.

The Genie's Take

The narrative summary, pre-generated at match time. Safe to share (uses aggregate stats only, no raw PII).

"Where the time went"

A stacked-bar visualization showing how much time each phase consumed (prep, blocking, scoring, classification, summary). Useful for diagnosing why a particular match took longer than expected.

"How the Genie matched"

Engine telemetry:

Confidence threshold used
Average composite score of matches
Training mode (if the match used a custom profile)
Blocking rules that were used
Pass breakdown (exact_id / deterministic / fuzzy / phonetic counts)

Analytics panel

Group-by breakdowns (match rate by state, by company, by data quality band). These are generated server-side and pulled in when the detail page loads.

If any review cases remain unresolved, they show here — same side-by-side comparison UI as the Review step. Resolve cases from here any time, even months after the match ran. Your accept/reject decisions persist across reloads.

Matches browser

A searchable, filterable table of every match the job produced — matched, review, and unmatched combined. Use it to spot-check without downloading Excel. Columns:

Source / Master row IDs — when the job ran in all_candidates mode, a source row that matched two masters (e.g. the same physician at two practice locations) shows up as two rows with the same source ID and different master IDs. Quickly see which cases are multi-practice vs single-match.
Status chips — All / Matched / Review / Unmatched with live counts.
Score range slider — filter to the borderline band (say, 70–80%) to focus review on the toughest cases.
Free-text search — matches any field value across source + master + match reason.
Sort by any column — click any column header.
Click a row to expand — full source record side-by-side with full master record, with agreement chips on paired fields (match / differs). The Genie's "Why matched" reason shows at the bottom.
Inline accept / reject — review-bucket rows have Accept / Reject buttons that save immediately. Same decisions API as the Review Queue, so changes round-trip.

The view icon stays visible; everything else lives in a dropdown menu to keep the list tidy:

Export results — downloads the results CSV for this specific job
Duplicate setup — opens the wizard on the Configure step with this job's files, column mapping, profile type, match mode, confidence, and ZIP radius settings all pre-filled. Perfect for "same files, different settings" iteration: re-run with all_candidates mode instead of best_match, swap one column to Ignore, enable ZIP radius, etc. No re-uploading or re-mapping.
Retry match — re-runs the job from scratch (e.g. after re-preparing a file).
Delete — below a divider so it's harder to misclick.

Export section

Four buttons: CSV, XLSX, PDF Report (Pro+), PPTX (Business). Exports are regenerated fresh from the stored job data — you always get the current state including any review-queue decisions you've made since the original export.

Deleting a job

Click Delete in the job header. You'll get a confirmation dialog that explains:

The job record will be deleted
The input files are not deleted (those remain on the Files page)
Shareable links for this job's report become invalid

Deletions can't be undone.

Dashboard — recent jobs overview
Reports — generate structured reports from any completed job
Review — the review UI (same UI is embedded in job detail)