← Course kit · Assessment kit · Day 2 · answer keys inside

Applied AI at TIYA Services: Day 2 Assessment Kit

Certificate criteria, the Day 2 knowledge check and exit checklist, answer keys for the Day 2 exercises, and the evaluation form. Print the participant pages without the keys.

Day 2Oct 21, 2026

Proposal · confirm with the program owner

Certificate criteria

The syllabus promises attendance certificates to participants who successfully complete the course. This kit defines "successfully complete" so it's about showing up and doing the work, not a test score.

Day 2

Agentic AI with Microsoft Copilot

  • Attended all blocks; up to 30 minutes' absence excused.
  • Built at least one agent and ran five tests on it.
  • Completed the governance card.

The knowledge check is formative. Participants score their own; instructors collect the anonymous scores per question to report which ideas didn't land.

Day 2

Exercise answer keys

Agent or not?

#AnswerWhy
1POne-time task.
2ARepeated questions, defined knowledge. The SOP Navigator.
3TNeeds a schedule or new-email trigger, so Copilot Studio or Power Automate, with a person reviewing the post.
4NA decision with pricing. AI can help organize non-price criteria; people decide. Expect debate.
5ARepeated, structured analysis: the Work Order Analyst. A saved prompt also works.
6NPayroll math belongs in the payroll system, which gets it exactly right every time.
7AKnowledge Q&A for a defined audience.
8POne-time draft.
9TTriggered, but safety-critical: the AI can flag, while the existing desk process still screens every request. Expect debate.
10TRules-based automation (Power Automate or the HR system). No generative AI needed.

Work Order Analyst: the figures to check

File: work-orders-aug-sep-2026.csv. Latest date in file: September 29, 2026.

  • Scope: 80 rows, opened August 2 to September 29, 2026.
  • By priority (all rows): Emergency 2, Urgent 21, Routine 37, Scheduled 20.
  • Status: 63 closed, 17 open (14 Routine, 3 Urgent; one of the Urgent is "On Hold - Parts").
  • Overdue Routine (>30 days), 9: 26-04306, 26-04314, 26-04321, 26-04325, 26-04329, 26-04335, 26-04339, 26-04342, 26-04344.
  • Overdue Urgent (>3 days), 3: 26-04373 (B-216 flush valve), 26-04374 (B-214 RTU-2 compressor, on hold), 26-04376 (B-101 breaker, server closet).
  • Repeat failure: B-214 RTU-2. Seven work orders from August 3 to September 22 (26-04303, -04322, -04332, -04345, -04359, -04370, -04374), including both Emergencies in the file, 21.5 labor hours.
  • The trap: B-410 Electrical also has 7 rows, but 3 are scheduled preventive maintenance and the rest are unrelated outlets and an exit sign. An agent that counts building and system combinations without reading descriptions will report it as a repeat failure. Use this in the debrief.

SOP Navigator tests

The expected answers are in the Day 2 workbook's test plan; they come from the sample SOPs. Common failure: test 4 answered from general knowledge. The fix is the "I can't find that in the SOPs" instruction, and moving it higher in the instructions if it's ignored.

Day 2 · 4:20 – 4:30

Knowledge check and exit checklist

  1. What makes an agent different from a prompt?

    • A. It uses a better model.
    • B. Standing instructions, its own knowledge, and tools for a defined job.
    • C. It can never be wrong.
    • D. It runs on its own server.

    B.

  2. What starts an agent built in agent builder?

    • A. A new email
    • B. A schedule
    • C. A person asking it something
    • D. Any of the above

    C. Events and schedules need Copilot Studio or Power Automate.

  3. You share your agent. A colleague gets "I can't find that" for a question your SharePoint document answers. The most likely cause?

    • A. The model is broken.
    • B. They don't have access to that document.
    • C. They asked too politely.
    • D. Agents only work for their creator.

    B. Sharing the agent doesn't share its knowledge sources.

  4. What's the strongest protection against an agent revealing personnel information?

    • A. An instruction saying "never reveal personnel information"
    • B. A polite description
    • C. Not putting personnel information in its knowledge
    • D. Telling users not to ask

    C. An instruction is a request; what the agent can't read, it can't leak.

  5. You change your agent's instructions. What next?

    • A. Nothing
    • B. Rerun your test set before re-sharing
    • C. Delete and rebuild it
    • D. Ask Copilot if the change is OK

    B.

  6. Which agent is tier 3?

    • A. A personal agent that formats your meeting notes
    • B. A team SOP Q&A agent
    • C. An agent that drafts responses to the COR
    • D. A personal brainstorming agent

    C. It drafts communication to the government.

Exit checklist (instructor initials each)

Day 2 · last 3 minutes

Course evaluation

1 = strongly disagree, 5 = strongly agree.

Statement12345
The exercises were relevant to my job.
I had enough hands-on time.
I know what data I can and can't put into Copilot.
I will use what I learned within the next week.
The pace was right.
The instructors answered my questions clearly.

The most useful thing today was:

One thing to change:

What would help you keep using AI at work?