Checkpoint: All checkpoints

A fictional client, for illustration. Release readiness for the claims-triage assistant. Historical week 4 of 6 snapshot.FERROWICK MUTUAL - ENGAGEMENT REPORTA fictional client, for illustration. Release readiness for the claims-triage assistant. Historical week 4 of 6 snapshot.2 October 2026 · Fictional client and data as of 2 October 2026; this historical sample was compiled later. Pick a checkpoint to follow it.KEY FINDINGWeek 4 of 6: 11 of 20 planned items done, 3 in progress, 1blocked on a client decision. Two decisions wait onFerrowick Mutual.Evidence: ev-items, ev-phasesFINDINGEngagement-wide decision waiting on Ferrowick Mutual: Whatblocks a release: a schema or policy failure only, or also any qualityregression over 2 points on the scorecard? Asked 2026-10-01.Evidence: ev-decisionsFINDINGEngagement-wide decision waiting on Ferrowick Mutual: Grantdata access to the region B claims extract so the 30 partial-liabilitycases can join the golden set? Asked 2026-09-26.Evidence: ev-decisions20Planned6 weekly checkpoints11Completed3In progress1BlockedAbout this sampleFerrowick Mutual is not a real company and nothing here describes one. Every engagement detail and evidencefile is fictional. The data depicts 2 October 2026; report editions record when the sample was compiled. ThisIGMD report was generated from a versioned sample tracker using the same build path intended for engagementdata.
A fictional client, for illustration. Release readiness for the claims-triage assistant. Historical week 4 of 6 snapshot.Engagement-wide now, waiting, next, and prioritiesItemCheckpointSinceNowCost gate: cost per accepted triage at or under 0.08 USD,…Gates2026-10-01NowCheckpoint: the three gates run on every build and block …Gates2026-10-02NowCalibrate the automated judge against the reviewer panel …Evals2026-10-01WaitingAdd 30 partial-liability cases from region B, which the b…Golden set2026-09-26NextStand up the reviewer panel: three adjusters, one hour a …EvalsplannedNextCheckpoint: judge agreement with reviewers is good enough…EvalsplannedNextAssemble the release evidence bundle: gate results, eval …ReleaseplannedNextGo or no-go, signed by the claims lead against the bundleReleaseplannedNextCheckpoint: the first production release carries its evid…ReleaseplannedPriority: highAdd 30 partial-liability cases from region B, which the b…Golden setblockedPriority: highCost gate: cost per accepted triage at or under 0.08 USD,…Gatesin progressPriority: criticalCheckpoint: the three gates run on every build and block …Gatesin progressPriority: mediumStand up the reviewer panel: three adjusters, one hour a …EvalstodoPriority: criticalAssemble the release evidence bundle: gate results, eval …ReleasetodoPriority: criticalGo or no-go, signed by the claims lead against the bundleReleasetodo[1]Completed work by checkpointBaseline3Golden set3Taxonomy3Gates2Evals0Release0[2]The work this checkpointWork itemStatusLast movedProofBaseline — Baseline - what breaks today3/3 done2026-09-123 artifactsTaxonomy — Failure taxonomy - deterministic or a matter of judgement3/3 done2026-09-193 artifactsGolden set — Golden set - replayable cases, frozen3/4 done, 1 blocked2026-09-263 artifactsGates — Deterministic gates - run on every build2/4 done, 2 in progress2026-09-303 artifactsEvals — Quality evals - the scorecard against the client's reviewers0/3 done, 1 in progress1 artifactRelease — Release gate - go or no-go, with the evidence attached0/3 done0 artifacts[3]
A fictional client, for illustration. Release readiness for the claims-triage assistant. Historical week 4 of 6 snapshot.The proof packet - where each claim can be checkedItemArtifactWhereBaseline3 artifacts across 3 itemsevidence/Taxonomy3 artifacts across 3 itemsevidence/Golden set3 artifacts across 4 itemsevidence/Gates3 artifacts across 4 itemsevidence/Evals1 artifact across 3 itemsevidence/Release0 artifacts across 3 itemsevidence/[4]Browse the fictional evidence filesOpen the sample evidence index to inspect the fabricated artifacts behind this report. Each file is labeledfictional.Checks that ran on the build, not in a conversationCheckResultWhenObservedgolden set replaypass2026-09-26120/120 cases replay identically on two machinesschema gatepass2026-09-30118/120 valid against intake schema v3; 2 real …policy gatepass2026-09-30120/120 within adjuster authority after the tab…cost gatewarn2026-10-010.11 USD per accepted triage; target 0.08; 60% …judge calibrationwarn2026-10-01agreement with reviewers 43/60 (72% rounded); t…[5]Decisions - answered and waitingDecisionAskedAnswerd1: Approve failure taxonomy v1 as the contract for regression …2026-09-18approve (Ferrowick Mutual, 20…d2: What blocks a release: a schema or policy failure only, or …2026-10-01waiting on Ferrowick Mutuald3: Grant data access to the region B claims extract so the 30 …2026-09-26waiting on Ferrowick Mutual[6]Engagement-wide risks, concerns, dependencies and issuesKindItemOwnerStatusRiskThe reviewer panel needs three adjusters for an hour a …Ferrowick Mutualopen · medium/highConcernThe 40 captured failures under-represent regional parti…Ferrowick Mutualopen, seen as f4DependencyAdd 30 partial-liability cases from region B, which the…Ferrowick Mutualblocked 2026-09-26IssueGates block the build: moved from 2026-10-02: the cost …Triuna Labsdue now 2026-10-09[7]
A fictional client, for illustration. Release readiness for the claims-triage assistant. Historical week 4 of 6 snapshot.Milestones - declared dates, and the one that movedMilestoneItems doneReachedDueBaseline agreed3/32026-09-122026-09-12Taxonomy approved3/32026-09-192026-09-19Golden set v1 frozen3/32026-09-262026-09-26Gates block the build2/4open2026-10-09 (was 2026-10-02)Judge trusted or demoted0/3open2026-10-16First release with evidence0/3open2026-10-23[8]The work log - items completed, cumulativeitems0358101311 Sep12 Sep17 Sep18 Sep19 Sep24 Sep25 Sep26 Sep30 SepItems completedWhen each milestone was reached12 SEPBaseline19 SEPTaxonomy26 SEPGolden setEditions of this reportEditionBuiltContent hashStatus at that editionv12026-10-24 22:12afc493d26dc1Week 4 of 6: 11 of 20 planned items done, 3 in progress, …v22026-10-24 22:18c3df8f2d0ef7Week 4 of 6: 11 of 20 planned items done, 3 in progress, …[9]Appendix - what this report rests onev-items · tracker.work_log · 26 rows × 4 columns (Work item, Status, Last moved, Proof)ev-phases · tracker.completed_by_phase · 6 rowsEngagement-wide decision waiting on Ferrowick Mutual: What blocks a release: a schema or policy failure only, or alsoany quality regression over 2 points on the scorecard? Asked 2026-10-01.ev-decisions · tracker.decisions · 9 rows × 3 columns (Decision, Asked, Answer)Engagement-wide decision waiting on Ferrowick Mutual: Grant data access to the region B claims extract so the 30partial-liability cases can join the golden set? Asked 2026-09-26.ev-decisions · tracker.decisions · 9 rows × 3 columns (Decision, Asked, Answer)
A fictional client, for illustration. Release readiness for the claims-triage assistant. Historical week 4 of 6 snapshot.SOURCES & EVIDENCEData as of 2 October 2026.tracker: data/ferrowick.json · from https://triunalabs.com/consulting/stories/ · sha256:b0ba8374f5b5...Evidence: ev-items = tracker.work_log; ev-proof = tracker.proof; ev-checks = tracker.evidence_packet; ev-decisions = tracker.decisions;ev-phases = tracker.completed_by_phase; ev-raid = tracker.raid; ev-milestones = tracker.milestones_tableNotes1. This is the whole engagement even when a checkpoint is selected. Now is in progress; waiting is blocked onsomeone; next is planned and not started; priorities are declared on items.2. One bar per weekly checkpoint. Pick a bar and the status, tiles, work, proof, checks, decisions and milestonefollow it. Engagement-wide priorities and risks remain visible.3. Under all checkpoints, one row each. Under a checkpoint, every item with its status, the day it last moved, andhow much proof it recorded.4. Every listed path resolves through the linked fictional evidence index. In a real engagement, files would stayunder the client's approved access controls.5. Replay, schema, policy, cost and judge calibration, each with what it was measured against.6. Every decision the tracker records, with its answer and who gave it, or who it waits on. No due date is shownbecause none was declared on a decision.7. This is the whole engagement even when a checkpoint is selected. Risks have likelihood and impact; concernsawait assessment; dependencies are blocked; issues record moved dates.8. Due dates are the ones declared at the start. A moved date shows the original beside it; the reason is in theissues row above.9. Each published edition records the hash of its content. The same pinned inputs rebuild the same bytes, so thereport can be checked against its source.