AgentOps Workshop
This site is for instructors preparing to deliver the AgentOps workshop. It packages the slides, the narrated walkthrough, the speaker script, and the run-of-show in one place.
Two tracks are available:
- AgentOps Briefing (~1 hour) - ready to deliver.
- AgentOps Value Delivery Workshop (~8 hours, Value-Based Delivery) - facilitator-ready, hands-on labs.
AgentOps Briefing - ready to deliver
Repository access required: this repository is private, so the preview and download links resolve only for signed-in members of the Azure GitHub organization.
| Resource | What it is |
|---|---|
| Datasheet (PDF) | One-page overview: audience, flow, and outcomes. |
| Delivery guide (PDF) | How to deliver the briefing, section by section. |
| Agenda | Customer-facing abstract and timeboxed agenda. |
| Run of show | Presenter timing, transitions, and delivery cues. |
| Speaker script | Verbatim word-for-word narration (~50 min). |
AgentOps Value Delivery Workshop
The full-day (~8 hours) workshop is the Value-Based Delivery (VBD) track. Across seven connected, hands-on labs, attendees take one agent - the Contoso Travel Agent - all the way around the operating loop on Microsoft Foundry. Every lab consumes the artifact the previous lab produced, so attendees leave with a working end-to-end AgentOps pipeline, not disconnected demos.
Repository access required: this repository is private, so the preview and download links resolve only for signed-in members of the Azure GitHub organization.
| Resource | What it is |
|---|---|
| Datasheet (PDF) | One-page overview: the seven labs, audience, and outcomes. |
| Delivery guide (PDF) | Room logistics, pacing checkpoints, and lab-by-lab notes. |
| Agenda | Full-day flow, audience, and facilitator checkpoints. |
| Lab roadmap | The seven labs, durations, and the continuity spine. |
| Lab guide | Step-by-step hands-on labs (Lab 1 through capstone). |
| Observability strategy | Trace, telemetry, and dashboard plan for the deep dive. |
What the workshop is about
AgentOps applies production engineering discipline to AI agents. The workshop walks instructors through the four-pillar operating model used to take one agent from prototype to production:
Evaluate → Ship → Observe → Operate
AgentOps Briefing introduces the model with a demo storyline: an existing Foundry agent is evaluated, release gates block regressions before production, runtime traces make behavior observable, and Day-2 operations turn incidents and production learnings into the next evaluation cycle.
Authoring this workshop
If you are not delivering the workshop but contributing to it, start with the repository README on GitHub. The README and the prep/ folder hold the authoring tools, references, and intermediate artefacts used to build everything published here.