Program agenda · Day 3

Advanced Agents with Copilot Studio

Applied AI at TIYA Services · Day 3. Agents that run without being asked. Participants build agents in Microsoft Copilot Studio that start on an incoming email, run on a schedule, and hand work to other agents, then learn to operate them safely: untrusted input, monitoring, cost, and a documented off switch. Every build ends with a person reviewing the agent's output. The day opens with a short review of Day 2.

Day 3
[Date TBD]
Hours
8:00 AM – 4:00 PM
Tool
Microsoft Copilot Studio
Location
TIYA Services Office
Instructors
Alex Djalali, Alex Rhodeen

Day 3 · [Date TBD] · 8:00 AM – 4:00 PM

Advanced Agents with Copilot Studio

8 AM910111212345 PM
Instruction Hands-on build Break / lunch Keyboard time 5.25 h

Morning · Always-On Agents

8:00 AM – 11:45 AM
8:00 – 8:30
Welcome & Day 2 in Review
  • Copilot Studio login check
  • The autonomy ladder: from on-demand agents to triggered and autonomous agents
  • What changes when no one is watching: untrusted input, real actions, real cost
8:30 – 9:45
From Agent Builder to Copilot Studio
  • A tour of Copilot Studio: knowledge, tools, agents, topics, triggers, and the activity view
  • Tools and connectors: how an agent acts in other systems
  • Build 1: the SOP Navigator, rebuilt, retested, and published for other agents to use
Hands-on build
9:45 – 10:00
Morning break
10:00 – 11:45
Event-Triggered Agents
  • Trigger, condition, payload, action, and human check
  • Agent instructions vs. payload instructions
  • Untrusted input: the email is data, not instructions
  • Build 2: a Work Order Intake agent that triages new request emails for a Facilities Lead
Hands-on build
11:45 – 12:30
Lunch

Afternoon · Agent Teams & Operations

12:30 PM – 4:00 PM
12:30 – 1:30
Scheduled Agents
  • Scheduled prompts in Copilot Chat vs. recurrence triggers in Copilot Studio
  • Let the flow count; let the agent write
  • Build 3: a Monday Backlog Briefing posted to Teams for a lead to review
Hands-on build
1:30 – 2:45
Multi-Agent Orchestration
  • Orchestrators and specialist agents
  • Child agents vs. connected agents, and when one agent is enough
  • Build 4: a Facilities Desk that routes questions to three specialist agents
Hands-on build
2:45 – 3:00
Afternoon break
3:00 – 3:45
Operating Autonomous Agents
  • Five new risks: injection, credentials, silent failure, cost, and chained errors
  • Red-teaming a triggered agent
  • Runbooks, risk tiers, and the off switch
Activity: Red-team & runbook
3:45 – 4:00
Closing & Assessment
  • Knowledge check, trigger shutdown check, and next steps

Program details

Format
A full-day, in-person session with a limited number of remote seats.
Prerequisite
Day 2, Agentic AI with Microsoft Copilot, and a Copilot Studio license.
Cohorts
About 30 participants per cohort. Cohorts attend on separate days for more individual attention.
Program lecturer
Alex Djalali
Instructor
Alex Rhodeen
Materials
Lecture materials, presentations, and hands-on exercises.
Room & AV
Classroom at the TIYA Services Office for 30 participants, with AV for presentations, internet access, and support for remote students.
Security requirements
None
Certificates
Each participant who completes the day receives a certificate of attendance.