Map the Claims Intake Before Automating It
Build an evidence-backed flow, identify the dominant delay and define a controlled future-state test.
22 minutes
Scenario
HarborSure wants AI to extract data from household claims. Customers wait days for a first decision even though intake activity takes minutes.
- Your role
- Operations analyst leading the current-state review
- Method
- Value Stream Mapping
Evidence pack
Demand
1,200 standard household claims arrive weekly; Monday volume is twice the daily average.
Intake
File registration takes 6 minutes and waits a median 4 hours.
Extraction
Manual field capture takes 9 minutes and is 94% complete on first pass.
Validation
Cases wait 19 hours; 27% lack one required loss document.
Decision
Standard review takes 8 minutes and waits 26 hours.
Rework
Missing-document requests add a median 2.4 days and are tracked in email.
Proposal
The vendor predicts 80% faster extraction but provides no end-to-end lead-time estimate.
Data
Timestamps exist in three systems and claim identifiers are inconsistent in 7% of records.
Constraints
- The first test may include only standard household claims from two document types.
- Low-confidence and missing-document cases require a visible human exception route.
- Customer documents may be processed only in the approved environment.
Case steps
Work through each prompt using the evidence pack. Answers and rubric weights stay protected in the interactive flow.
Define the customer outcome, demand family, start and end, then arrange e1-e8 into a current-state work and information flow.
Identify the dominant barriers to flow and explain why extraction speed alone is insufficient.
Design a future-state normal path and exception path, including the role and boundary of AI extraction.
Define a bounded implementation test with owner, measures and a scale decision rule.