AIBarrysecure workshop studio

for the room you actually have.

Choose the audience and available time. Get a complete facilitator runbook built on the AI Security Learning Journey.

01 · Audience

Who is in the room?

End every section with the CISO decision, developer action, and administrator control.

02 · Time

How much time do you have?

Every option has an exact agenda total and a deliberate stopping point.

Mixed Team2 hours

AudienceMixed Team

Total time

Agenda steps

Ready resources

Why this plan

Designed around what Mixed Teams need to leave with.

Build a shared view of AI security risk.
Connect executive decisions to engineering and operations.
Leave with named owners and a common next step.

Before people arrive

Facilitator preparation

Ask attendees to identify as decision maker, builder, operator, or more than one role.
Open every selected resource and verify the Demo Runner and first lab before attendees arrive.
Arrange mixed-role groups for hands-on and debrief activities.
Explain that every debrief ends with a leader decision, developer action, and administrator control.

Run of show

Follow the room, not a script.

Use the current-step controls during delivery. Every section includes the objective, facilitator moves, participant action, and recovery path.

YouTube introVideo coming soonThe agenda is ready now. Add the video URL later without changing the page.

Say · show · do

Say: AI security spans applications, models, agents, tools, and the systems around them.
Show the intro video when available; until then, use the linked video collection as optional support.
Ask attendees to note one risk that already exists in their environment.

Participant action

Note one leadership decision, engineering question, or operational gap.

Expected result

Attendees can describe why AI risk is different from a conventional application-only threat model.

Debrief prompt
What must leaders decide, developers change, and administrators operate?

Transition

Use the observations to choose the most relevant risk and use-case examples.

If time is short

Use a two-minute verbal framing and share the video collection for follow-up.

Journey previewAI Defense videosCollection of AI Defense and AI security presentations and demonstrations.Open resource

Say · show · do

Say: application security protects prompts, responses, models, and data paths; agentic security also governs tools, skills, identity, and action.
Show one request path and one agent-tool path, naming the control points on each.
Ask the group who currently owns each control point and where ownership is unclear.

Participant action

Map one use case across business owner, development owner, and platform owner.

Expected result

The group can distinguish the two motions and identify at least one ownership gap.

Debrief prompt
Where would a risky request or agent action be detected today?

Transition

Move from the architecture to a live scenario in the Demo Runner.

If time is short

Use one application example and one agent example; move additional cases to follow-up.

◇30-minute overview deckResource link coming soon

Say · show · do

Open the Demo Runner before speaking so the scenario is ready.
Run one safe request and one adversarial request, narrating what reaches the model and what is stopped.
Point out the decision, evidence, and control location without getting lost in setup details.

Participant action

Identify the business consequence, engineering control, and operational evidence after each run.

Expected result

Attendees see a concrete difference between an unprotected path and an enforced control.

Debrief prompt
What evidence would your team need to trust and operate this control?

Transition

Use the demonstrated failure mode to introduce the primary exercise.

If time is short

Run one adversarial scenario and describe the safe baseline verbally.

Journey previewAI Defense Demo RunnerRun guided AI Defense quick-start demos and walkthrough scenarios from a single launcher.Open resource

Say · show · do

Confirm everyone reaches the CTF before starting the timer.
Explain the first challenge and let participants work independently or in pairs.
At checkpoints, ask what failed, what evidence appeared, and where a guardrail belongs.

Participant action

Complete the assigned challenges and record one control recommendation per finding.

Expected result

Participants can explain at least one successful attack and a concrete mitigation.

Debrief prompt
Which control would have prevented the attack earliest and with the least friction?

Transition

Carry the attack lessons into the next application or agent exercise.

If time is short

Assign the first challenge only and provide the remaining CTF as self-paced follow-up.

Journey previewAI Defense CTFHands-on Capture The Flag challenges to test your AI security skills in real-world scenarios.Open lab

Say · show · do

Ask what happened before explaining the intended lesson.
Separate the root cause from the visible symptom.
Capture the earliest useful control and the person who should own it.

Participant action

Share one observation and one change they would make in a real environment.

Expected result

The exercise produces a specific lesson and next action.

Debrief prompt
What is the CISO decision, developer action, and administrator control?

Transition

Connect the lesson to the assessment and broader learning journey.

If time is short

Take one observation and one action from the room.

◎AI Security AssessmentTurn workshop observations into a maturity baseline and action plan.

Say · show · do

Recap the most important risk, control, and ownership insight.
Show the assessment and complete learning journey as the two continuation paths.
Confirm the next action, owner, and date before ending.

Participant action

State one action they will take and bookmark the relevant follow-up resource.

Expected result

The workshop ends with an owned next step and accessible continuation resources.

Debrief prompt
Who will do what next, and when will the group review progress?

Transition

End the live session and leave the self-paced resources available.

If time is short

Confirm one owner and one follow-up date, then share the remaining links.

Current1 of 6

Continue after the room clears

Keep the momentum.

Labs remain self-paced. Send the journey, capture the assessment, and agree when the team will review progress.

↗Complete AI Security Learning JourneyThe full self-paced collection behind this workshop planner.◎AI Security AssessmentTurn workshop observations into a maturity baseline and action plan.Journey previewAI Defense videosCollection of AI Defense and AI security presentations and demonstrations.Open resource

for the room you actually have.

Who is in the room?

How much time do you have?