What did the AI estimate leave unproved?
A diagram and price can omit both a retained cost and the customer data boundary.
The question
Larkspur receives an authored AI-generated estimate and diagram. The customer requires full source records to remain in its named region. All AI text, people and results are fictional teaching material.
The new target is ready at $11,000 per period. The app is drawn inside the customer’s approved region.
- Finance$9,000 of old environment cost remains contracted for 3 periods.
- Data ownerFull source records must remain in the named region. No acceptance of this proposed processing path has been supplied.
“Identify what the estimate establishes, what it omits and who can accept the remaining claims.”
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: What do you inspect before presenting the estimate as a commitment?
What would you do here? Open the move you’d make.
Present the diagram as validation
Tempting, but it misses something. A proposed artifact is not an observed result.
Inspect inputs, retained costs and data paths
The strong move. This can expose the claims the diagram has not established.
Ask for a confidence percentage
Tempting, but it misses something. A percentage does not supply the missing facts.
See how a strong candidate says it
“Inspect the model inputs, retained costs and the actual data paths. The proposed diagram is not evidence that every processing location or recurring cost is included.”
Why it works: An estimate can help explore assumptions without proving them.
Step 2: The processing path sends full records outside the named region. What changes?
What would you do here? Open the move you’d make.
Trace and bound the full-record path with its owner
The strong move. This responds to the actual requirement conflict.
Call the architecture accepted because the app is in-region
Tempting, but it misses something. The displayed processing path contradicts that claim.
Assume a generally available tool has customer approval
Tempting, but it misses something. No customer approval was supplied.
See how a strong candidate says it
“The design conflicts with the customer’s stated requirement. Trace and bound that path with the data owner before proceeding. The app’s drawn location does not establish every processing location.”
Why it works: The requirement concerns record handling, not the placement of one box.
Step 3: A sample-data demonstration completes a happy path. What does it prove?
What would you do here? Open the move you’d make.
Treat the demo as production equivalence
Tempting, but it misses something. The customer’s representative contracts and data were not tested.
Ban all tools permanently
Tempting, but it misses something. The missing evidence does not establish that every tool is unusable.
Record its scope and define the missing checks
The strong move. This keeps a useful demonstration from becoming an unsupported acceptance claim.
See how a strong candidate says it
“It shows that the exercised sample path ran. It does not establish representative contract behavior, recovery or the customer’s data boundary. I would name those missing checks and their owners.”
Why it works: Demonstration evidence has a specific scope.
Step 4: The data owner accepts only a restricted projection. What may you claim?
What would you do here? Open the move you’d make.
Record acceptance for the whole architecture
Tempting, but it misses something. That extends a narrow decision to unaccepted scope.
Revise the boundary and retain the remaining evidence
The strong move. This uses the actual acceptance without manufacturing a larger one.
Hide the retained environment cost
Tempting, but it misses something. The transition budget remains incomplete.
See how a strong candidate says it
“That narrow acceptance can support a revised projection boundary. It does not approve the full-source move, complete operating acceptance or fix the omitted retained cost.”
Why it works: Acceptance must stay attached to the scope that was accepted.
The answer, in short
- Decision
- Revise the data boundary and estimate; do not present the proposal as accepted.
- Because
- The full-record path conflicts with the requirement, the price omits retained cost and the demonstration has limited scope.
- What would change it
- Specific data-owner acceptance, representative technical evidence and a complete transition model can support a new commitment.
Common traps
- Turning a generated artifact into evidence.
- Confusing app location with every data path.
- Extending a narrow acceptance to the whole design.
Check your answer
- Did I inspect the estimate’s inputs?
- Did I trace the record boundary?
- Did I bound the demonstration claim?
- Did I keep acceptance attached to its actual scope?
Last reviewed Nov 10, 2026