GlanzsteinGenAI Demo Navigator
Landscape
Zoom-in · Repair-to-Delight

Repair-to-Delight — The Aftercare Desk

Swimlane view of Repair-to-Delight — click a step for its full detail; back to the landscape.

Overlays:
Customer & boutique
Intake & dossier agents
Entitlement & goodwill
Workshop & components
Delivery & comms
1Piece handed in 1 1
2Dossier assembly & piece identification 1 1
3Missing-info loop 1 1
4Entitlement verdict 1 1
5Quote or goodwill decision 1 2
6Component resolution & reservation 1 1
7Workshop repair & QC 1
8Return delivery 1
9Proactive status & aftercare comms 1 1
React Flow
Press enter or space to select a node. You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.
Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

Issue → AI potential map

Derived from this view’s pain points and potentials — the operating concept column is the production thinking.

IssueImpactAI patternOperating conceptKPI effectDemo
Repair intake loops for weeks before a technician sees the piecespeed, cx, costcycle time is dominated by intake completeness, not repair effortdocument-intelligenceExtraction is evidence-cited: every fact points at the photo, receipt line or registry record it came from, and the piece identification is confirmed against the piece registry rather than asserted. Incomplete dossiers trigger one precise request, not a correspondence.
  • First-pass intake completeness ↑
  • Repair cycle time ↓
Glanzstein Atelier — Agentic Repair & Aftercare Desklive
Repair intake loops for weeks before a technician sees the piecespeed, cx, costcycle time is dominated by intake completeness, not repair effortoutbound-agentMessages are generated from case facts and templates under a named sender; deadline watchdogs escalate a stalled case before the promise date is breached rather than apologising after. What leaves the house is logged with the state that produced it.
  • Status-chasing contacts ↓
  • Promise-date breaches ↓
Glanzstein Atelier — Agentic Repair & Aftercare Desklive
Entitlement is read from three overlapping layersquality, risk, costinconsistency is both a margin leak and a fairness problemvalidation-serviceThe agent must call the entitlement service and quote its verdict; a case cannot advance on a coverage judgement the model produced. This is the retail analog of the telco offer engine — the deterministic authority the language model defers to. Goodwill above the published cap routes to a human, every time.
  • Entitlement decision consistency ↑
  • Goodwill leakage ↓
Glanzstein Atelier — Agentic Repair & Aftercare Desklive
Matching a retired piece to a serviceable component is expert lorespeed, quality, risktribal knowledge is a single point of failure for the aftercare promiseagent-assistEquivalence chains are data, not model memory: the agent reads the component system's mappings and quotes them, and a missing mapping is a question for the atelier veteran — captured back into the chain once answered, so the lore compounds instead of retiring with its keeper.
  • Component identification time ↓
  • Repairs blocked on parts ↓
Glanzstein Atelier — Agentic Repair & Aftercare Desklive

Figures marked “actual” come from FY2025 public reporting; ranges marked “illustrative” are industry-plausible benchmarks, not company data.

GenAI demo platform — company alias content; figures partly illustrative (see basis flags). Privacy