Microsoft AB-100 Practice Test 10

Question 90

A tuned custom classifier improves invoice routing accuracy slightly but triples support and monitoring cost. How should the architect decide whether to use it?

Compare the measured incremental business benefit against full model lifecycle cost and risk.
Use only development training loss rather than held-out business outcomes.
Ignore data monitoring because tuning happened once.
Assume any accuracy improvement justifies unlimited cost.

Explanation

Correct Answer

Custom-model economics depend on real error reductions, maintenance and retraining burden.

Incorrect Answers

Training fit is not proof of deployment benefit.

Model performance can change as invoices evolve.

Marginal gains may not justify large ongoing costs.

Question 91

A reusable prompt library contains an unsupported action name after Copilot Studio tooling changes. What should maintainers do?

Version the prompt, update its tool contract and rerun scenario tests before promoting it.
Keep the outdated action reference because prompts are only documentation.
Silently change production text without versioning or rollback.
Remove all validations and let the agent guess the intended tool.

Explanation

Correct Answer

Prompts that invoke tools are executable design assets and need compatibility and release checks.

Incorrect Answers

Broken action names can cause real workflow failure.

Untracked changes prevent incident reconstruction.

Tool guessing is not a safe replacement for explicit contracts.

Question 92

A manufacturer proposes an agent that saves invoice-processing hours but requires storage, model tokens, monitoring and exception reviewers. Which ROI model is defensible?

Include only the model API fee and ignore human exceptions.
Compare verified labor and error reductions against implementation and all recurring operating costs.
Treat the number of completed invoice pages as the sole benefit measure while excluding labor exceptions and recurrent costs.
Treat purchase price as the entire ownership cost regardless of operation.

Explanation

Correct Answer

A meaningful ROI uses a baseline plus total cost of ownership, not just the most visible cloud fee.

Incorrect Answers

Omitting ongoing reviewers and support makes benefits artificially high.

Page volume cannot be monetized without validated time saved, defect reduction and full total cost of ownership.

One-time purchase cost misses recurring operations and governance expense.

Question 93

An insurer processes 1,200 claims monthly. An AI assistant saves six analyst minutes per claim; analyst time costs $30/hour. Monthly AI licensing and monitoring cost $1,100. What is the expected net monthly saving?

$1,100 per month, counting only platform expenses.
$3,600 per month, ignoring running costs.
$13,000 per month based on mixing hours and dollars.
$2,500 per month.

Explanation

Correct Answer

The time saved is 120 hours monthly, worth $3,600; subtract $1,100 to obtain the $2,500 net.

Incorrect Answers

Platform expense is a cost, not the resulting net labor benefit.

That is the gross labor saving before platform expenses.

Unit consistency requires converting minutes to hours before multiplying by hourly cost.

Question 94

For a three-year AI service, custom build costs $120,000 initially plus $30,000 support annually; a prebuilt option costs $15,000 integration plus $55,000 licensing annually. Which is cheaper over three years?

Custom build by $30,000.
The options tie at $180,000.
Custom build by $120,000 because only the setup fee counts.
The prebuilt option by $30,000.

Explanation

Correct Answer

Build TCO is $120,000 + 3×$30,000 = $210,000; buy TCO is $15,000 + 3×$55,000 = $180,000.

Incorrect Answers

This reverses the actual totals and ignores the prebuilt recurring license arithmetic.

Only the prebuilt total equals $180,000, not the custom total.

Ignoring support understates lifecycle costs and does not produce the requested comparison.

Question 95

A company already has a supported Copilot Studio connector that meets an expense-approval requirement, but leadership proposes a costly custom app. What makes extending the existing component preferable?

It satisfies the verified capability and authorization requirements with lower implementation and support burden.
It permits skipping any change approval because it is prebuilt.
It automatically provides administrator rights to finance data.
It guarantees every task succeeds without monitoring.

Explanation

Correct Answer

Build/buy/extend decisions should compare real capability, risk and total lifecycle cost.

Incorrect Answers

Prebuilt components still need configured governance.

Connector access remains limited by verified identity and permissions.

Supported tools still need testing and telemetry.

Question 96

A service desk wants a custom-tuned model because its assistant sometimes cites the wrong policy section. Retrieval logs show an obsolete index and no source filter. What should be evaluated first?

Tune sampling temperature before testing whether the assistant retrieved the current approved warranty policy.
Train immediately on every historical unapproved policy.
Double GPU capacity without testing retrieval quality.
Repair source freshness and retrieval selection, then benchmark whether model tuning still adds value.

Explanation

Correct Answer

Grounding errors are often data or retrieval faults; expensive tuning is not the first remedy.

Incorrect Answers

Randomness cannot repair stale or wrongly selected source passages; retrieval must be validated before model tuning.

Training on obsolete policies can make the problem persistent.

Compute scale does not correct indexing or access filters.

Question 97

An AI routing service receives 10,000 routine requests at $0.001 each and 2,000 complex requests at $0.025 each. Sending all 12,000 through the costly model would cost $0.025 each. How much per-period model cost does routing save?

$60, which is the routed total, not the saving.
$25, by subtracting the per-request prices without volumes.
$240.
$300, which is the all-expensive cost, not the saving.

Explanation

Correct Answer

A routed mix costs $10 + $50 = $60; all-expensive costs $300, so the difference is $240.

Incorrect Answers

The routed expense is one side of the comparison.

Cost savings require applying prices to their respective request volumes.

The unoptimized expense is the baseline, not the benefit.

Question 98

A company calculates AI costs using inference tokens only. New users also need premium licenses, human review and integration support. Which correction makes the budget credible?

Include usage, licensing, people, data movement and operational support within the evaluation horizon.
Treat all human review as zero-cost because staff already work there.
Remove licensing from forecasts because it is paid separately.
Assume each request uses identical tokens despite varying document size.

Explanation

Correct Answer

Total ownership includes actual incremental and ongoing expenses and realistic workload distributions.

Incorrect Answers

Existing staff time still has opportunity cost.

Budget separation does not eliminate expense.

Variable context and output sizes make fixed per-request token assumptions unreliable.

Question 99

A bank evaluates an AI onboarding agent. The sponsor reports “50% faster” but does not know how many errors it creates. What ROI input is missing?

Value the 50% processing-time reduction at the full analyst rate without deducting cases that need extra manual correction.
Use the vendor-reported benchmark accuracy for a different onboarding workflow instead of measuring the bank’s own exception and rework rates.
Measure only user satisfaction with the onboarding assistant’s wording while omitting rework, errors and actual transaction completion.
A verified before-and-after baseline of completion quality, rework and exception handling cost.

Explanation

Correct Answer

Speed is only useful if failure, risk and rework costs are included in the business value calculation.

Incorrect Answers

Gross time saved overstates business return if higher error rates create additional review, rework or loss that offset productivity gains.

A vendor benchmark is not a comparable baseline for this bank’s data, controls and error cost. Actual onboarding outcomes are required.

A polished experience is not equivalent to verified workflow success or positive return on investment.

Question 100

A document assistant saves 600 analyst hours annually at $40 per hour. The annual subscription costs $9,000, and executives require at least $10,000 in annual net benefit. Ignoring one-time setup, what is the maximum annual human-review expense the project can support?

$15,000, treating gross savings minus the target as the reviewer budget and omitting subscription expense.
$14,000, treating total operating expense capacity as though all of it could be paid to reviewers.
$5,000 per year.
$24,000, treating all gross labor savings as available for review expense despite the net-benefit target.

Explanation

Correct Answer

Gross annual value is 600×$40 = $24,000. Required net benefit is $10,000 and subscription $9,000; remaining reviewer budget is $5,000.

Incorrect Answers

The $24,000 gross labor benefit must first fund the $9,000 subscription and leave $10,000 of required net benefit; $15,000 ignores licensing.

The total allowable operating spend is $14,000, but $9,000 is already committed to the subscription.

Using every dollar of savings for review expense leaves no room for licensing or the required retained net benefit.

Agentic AI Business Solutions Architect (AB-100) Practice Tests

Microsoft AB-100 Practice Test 8 - 10 Questions (Last Reviewed 06-Nov-2026)
Microsoft AB-100 Practice Test 9 - 10 Questions (Last Reviewed 05-Nov-2026)
Microsoft AB-100 Practice Test 10 - 10 Questions (Last Reviewed 04-Nov-2026)
Microsoft AB-100 Practice Test 11 - 10 Questions (Last Reviewed 03-Nov-2026)
Microsoft AB-100 Practice Test 12 - 10 Questions (Last Reviewed 02-Nov-2026)

Pass Your Microsoft AB-100 Exam Easy!

Question 90

Explanation

Correct Answer

Incorrect Answers

Question 91

Explanation

Correct Answer

Incorrect Answers

Question 92

Explanation

Correct Answer

Incorrect Answers

Question 93

Explanation

Correct Answer

Incorrect Answers

Question 94

Explanation

Correct Answer

Incorrect Answers

Question 95

Explanation

Correct Answer

Incorrect Answers

Question 96

Explanation

Correct Answer

Incorrect Answers

Question 97

Explanation

Correct Answer

Incorrect Answers

Question 98

Explanation

Correct Answer

Incorrect Answers

Question 99

Explanation

Correct Answer

Incorrect Answers

Question 100

Explanation

Correct Answer

Incorrect Answers

Agentic AI Business Solutions Architect (AB-100) Practice Tests