Operations director
Wants one risk queue and a measurable reduction in investigation time.
MISSION 01 · FLAGSHIP FDE CASE
FLAGSHIP LIVE
Atlas
Freight
Build a maritime delay decision system that reconciles unstable operational evidence and exposes financial risk before the decision window closes.
01 / CUSTOMER BRIEF
A working pilot exists. Nobody trusts the risk queue.
Atlas Freight coordinates maritime shipments for industrial customers. A dispatcher may need 42 minutes to determine whether a vessel delay threatens a contractual delivery window and creates financial exposure.
The company has already assembled a dashboard prototype. It displays vessel positions and port events, but users continue to investigate in spreadsheets, email, and carrier portals because the system cannot explain disagreements between sources or show which data is stale.
Your mission: design the minimum deployable workflow that reduces median investigation time below ten minutes while preserving uncertainty and human authority.
02 / STAKEHOLDERS
Four definitions of success.
The design must resolve real differences in incentives rather than averaging them into a generic requirement list.
Senior dispatcher
Knows the exceptions and fears the system will hide uncertainty.
Data platform lead
Wants a canonical model, stable contracts, and no direct load on production systems.
Finance partner
Requires explainable exposure calculations before discussing avoided cost.
03 / EVIDENCE ROOM
Partial truth across eight sources.
Use these evidence descriptions to create your own sample data and documents. A later release can add downloadable CSV, JSON, and interview files.
Dispatchers describe five different definitions of “delay” depending on the downstream decision.
Near-real-time positions with occasional gaps, duplicate pings, and changing vessel names.
Estimated and actual arrival/departure events using a different port code system.
Customer bookings and contractual windows, refreshed twice daily from a spreadsheet export.
Demurrage and operational exposure logic maintained by one analyst with undocumented overrides.
Messages showing how teams currently resolve disagreements between feeds and customer commitments.
A partial mapping among IMO numbers, internal vessel IDs, names, and carrier references.
Customer data cannot appear in the first pilot; production access requires separate review.
04 / TASK BOARD
Complete the deployment record.
Progress is stored locally in your browser. The sequence intentionally moves from decision and evidence to architecture, release, adoption, and value.
05 / DECISION CHECKPOINTS
Commit before the next fact arrives.
The mission is divided into reviewable checkpoints so the final solution cannot be reverse-engineered from the complication.
Discovery record
Submit the decision statement, stakeholder map, current workflow, baseline, constraints, and unresolved questions.
Gate: problem framingSolution and identity design
Submit the minimum deployable workflow, canonical model, source contracts, architecture, and risk register.
Gate: technical judgmentPilot readiness
Submit the working slice, failure behavior, rollout plan, training, rollback, and measurement instrumentation.
Gate: operational credibilityFinal recommendation
Revise after the complication and pilot evidence, then recommend scale, revision, continued pilot, or stop.
Gate: value judgment06 / FIELD COMPLICATION
Open only after your first architecture decision.
A good response does not throw away all prior work. It identifies which assumptions break, which artifacts must change, and how the rollout can be protected.
Reveal the Atlas Freight complication
Two weeks before the pilot, the primary vessel-position provider changes its identifier policy. Historical records retain the old internal identifiers, new records use a different provider key, and vessel names are not unique. The vendor will not provide a complete backfill before launch.
Required response: revise the identity model, define confidence and exception states, decide whether the pilot scope changes, and explain the effect on baseline comparability and risk calculations.
07 / SIMULATED PILOT REVIEW
Usage improved. The outcome is mixed.
After revising the identity design, evaluate the pilot evidence and decide what should happen next.
of invited dispatchers used the risk queue at least three days per week.
Median investigation time improved but missed the ten-minute target.
Overrides cluster around unmatched vessel identities and stale booking exports.
Alerts overstate risk when cost assumptions are incomplete.
“The queue gets me to the right shipment faster, but I still need the spreadsheet to understand why finance disagrees.”
Identity exceptions are visible, but the manual crosswalk process has no durable owner.
Expand to all routes next month and add automated customer notifications.
FINAL DECISIONRecommend scale, revise, continue the pilot, narrow the scope, or stop—and state what evidence would change your decision.
08 / SUBMISSION
Submit evidence of the whole deployment.
The strongest submissions are not the largest systems. They are coherent, explicit about uncertainty, and tied to the operational decision.
01Discovery record
02Stakeholder map
03Current-state workflow
04Problem statement
05Decision map
06Solution thesis
07Requirements and acceptance criteria
08Architecture diagram
09Data contract and identity policy
10Risk and control register
11Working solution or technical prototype
12Pilot charter
13Operating runbook
14Adoption plan
15Value measurement plan
16Executive recommendation
17AI-use disclosure
Mission scoring
| Dimension | What strong work demonstrates | Weight |
|---|---|---|
| Problem frame | Decision statement, workflow map, stakeholders, baseline, assumptions, and scope. | 15% |
| Data and identity design | Canonical model, source profile, matching rules, confidence states, and contracts. | 20% |
| Deployable workflow | Architecture or implementation covering interface, services, access, and failures. | 20% |
| Pilot and adoption | Users, training, support, overrides, escalation, rollout, and rollback. | 15% |
| Measurement | Technical, adoption, operational, and financial measures tied to the baseline. | 15% |
| Executive recommendation | Scale, revise, or stop—with evidence, uncertainty, risks, and next actions. | 15% |