← Course kit · Day 2 outline · for the program owner

Day 2 Outline: Agentic AI with Microsoft Copilot

The second day of Applied AI at TIYA Services. Day 1 taught participants to prompt Copilot safely. Day 2 teaches them to build no-code agents with the Copilot agent builder: standing instructions, their own knowledge, and tools for one defined job. Participants test each agent before sharing it, and govern it after.

Day 2Wed, Oct 21, 2026 Hours8:30 AM – 4:30 PM ToolCopilot agent builder InstructorsAlex Djalali, Alex Rhodeen

1

Audience and prerequisites

  • Who: TIYA staff in operational and administrative roles who completed Day 1. About 30 per cohort, with limited remote seats.
  • Prerequisite: Day 1, Generative AI for Daily Work. The current cohort took Day 1 in June 2026. Day 2 opens with a 10-minute Day 1 review, and nothing later depends on a Day 1 exercise or sample.
  • Accounts: a Microsoft 365 Copilot license, or at least Copilot Chat, with the agent builder enabled (section 5).

2

Learning outcomes

On completing Day 2, participants will be able to:

OutcomePracticed in
LO1Understand the fundamentals of AI agent creation and common agent architectures.Agent or not? · Design canvas
LO2Design, configure, and deploy autonomous agents using no-code and low-code tools.Builds 1 and 2 · Build your own
LO3Integrate reasoning, planning, tools, and memory into practical agent workflows.Build 2 · Trigger storyboard
LO4Apply security, privacy, safety, and trust principles when developing and deploying agentic AI.Red-team swap · Risk tier · Governance card

3

The day, 8:30 to 4:30

About 3½ hours of building, plus design and red-team exercises. Every agent is tested before it's shared.

TimeBlockTopicsActivityLO
8:30 – 10:15Agentic AI Foundations
105 min
Login check · Day 1 in review: Mask → Ask → Check and the data traffic light · from chat to agent: the autonomy ladder · anatomy of an agent · the no-code ecosystem and four architectures"Agent or not?" sort of ten TIYA tasks · agent design canvas for the participant's own workLO1 · LO4
10:15 – 10:30Morning break
10:30 – 12:15Building Real Agents
105 min · hands-on
Prompt engineering for agents: the instructions template · tests first · knowledge Q&A and multi-tool agentsBuild 1 SOP Navigator · Build 2 Work Order Analyst with code interpreter, numbers checked by handLO2 · LO3
12:15 – 1:15Lunch
1:15 – 3:00Building Real Agents, Continued
105 min · hands-on
Event triggers as a design concept, with an instructor demo · memory · publishing: permissions, naming, sharingTrigger storyboard · build your own agent from the morning's canvas · partner swap test · showcaseLO2 · LO3 · LO4
3:00 – 3:15Afternoon break
3:15 – 4:30Governance & Security
75 min
PII handling in agents: instructions are a request, permissions are a lock · risk tiers · the agent lifecycle and model governanceRed-team swap with five attack cards · risk tier · governance card · exit checklist, evaluation, and certificates from 4:20LO4

4

The builds

Two guided builds on fictional TIYA SOPs and work order data, then the participant's own agent. Step-by-step instructions are in the Day 2 workbook.

Build 1 · 10:45

SOP Navigator

  • Architecture: knowledge Q&A
  • Knowledge: the four sandbox SOPs
  • Teaches: the instructions template, tests first, "I can't find that" over guessing
Build 2 · 11:25

Work Order Analyst

  • Architecture: multi-tool
  • Knowledge and tools: FM-SOP-03 plus code interpreter, with an 80-row CSV uploaded in chat
  • Teaches: analysis with tools, and checking the numbers by hand
2:00

Build your own

  • From: the participant's design canvas
  • Teaches: five tests before building, partner swap test, safe publishing
  • Then: red-teamed and governed in the last block

5

Environment requirements

Confirm each item with TIYA IT, in writing, at least 14 days before.

If Copilot is down or blocked

The lead walks through saved build screenshots on screen, and pairs complete the design canvas and test plan on paper.

6

Assessment and certificate

  • Formative: a six-question knowledge check, self-scored. Instructors collect anonymous per-question scores.
  • Evidence of work: the test plan, the partner's swap-test log, the risk tier, and the governance card.
  • Certificate (proposed): full attendance (up to 30 minutes excused), at least one agent built and tested with five tests, and a completed governance card.

7

Materials

ItemFor
Day 2 slides (HTML, PowerPoint, PDF)Instructors · participants
Day 2 workbook: Day 1 review, autonomy ladder, design canvas, two builds, test plans, governance worksheetsParticipants
Day 2 agendaEveryone
Day 2 assessment kit: certificate criteria, answer keys, knowledge check, exit checklist, evaluationInstructors
Facilitator guide, Day 2 section: run sheet, talking points, demos, watch-fors, debriefsInstructors
The four sandbox SOPs (Word, and on one web page)Participants
Work order export, Aug–Sep 2026 (CSV, 80 rows)Participants

8

Decisions for the program owner

  1. Outcome 2 vs. the agent builder. LO2 promises autonomous agents and the agenda lists event triggers. The agent builder makes on-demand agents; triggers and autonomous agents need Copilot Studio, which is Day 3. Day 2 teaches triggers as a design concept with an instructor demo. Either soften LO2 ("design, configure, and deploy agents, and design event-triggered automations") or provide a Copilot Studio environment for the demo.
  2. Closing. The agenda ends at 4:30 with no closing block. The exit checklist, evaluation, and certificates run from 4:20 inside the governance block. If the day can run to 5:00, move them to a 4:30–5:00 closing.
  3. Session headers. The agenda's sessions are titled only "Morning" and "Afternoon". Suggested: "Agent Foundations & First Builds" and "Deploy, Trigger & Govern".
  4. Certificate criteria. Confirm the proposal in section 6.