Engineering & data · About 2 hours
Agentic Engineering
Build a small tool-using workflow with clear boundaries, observable steps and human review.
Your practical result
A constrained agent in a local lab, an execution trace and a small evaluation set with recorded results.
The outline below is built around this result. Final tools and setup instructions will be confirmed with the session details before registration opens.
Who this is for
Developers who can work with APIs and tests and want a focused introduction to agent workflows.
What you’ll practice
- Define a narrow task and a tool contract with explicit boundaries.
- Add observable steps, failure handling and review before a consequential action.
- Evaluate normal, ambiguous and disallowed requests against expected behavior.
Your two-hour workshop outline
A guided demonstration, a substantial build exercise, and time to check the result. Times are approximate and include discussion.
- 0–10 min
Constrain the task
Define a fictional service-request assistant and the actions it is allowed to propose.
- 10–30 min
Trace an agent workflow
Inspect tool input, tool output, decision points and a human approval boundary.
- 30–45 min
Define the tool contract
Specify allowed fields, rejected requests and an explicit stopping condition.
- 45–85 min
Build the agent
Implement a narrow workflow against sandboxed tools and capture its execution trace.
- 85–105 min
Evaluate boundary cases
Try a valid request, missing information, a tool failure and an unauthorized action.
- 105–120 min
Record the limits
Save evaluation results and identify what would need review before any production use.
Before you attend
- Comfort with programming, APIs, structured data and automated tests.
- Complete the starter-lab check. Use synthetic requests and sandboxed tools.
Tools and access
The framework, model account and any API-credit requirements will be confirmed before registration. The exercise does not require connecting a live customer system.
Delivery platform, location or joining instructions will be published with each confirmed session. Use practice material you have permission to share.
Your instructor
Elliott Lamboy
Elliott teaches these Texas AI Consulting workshops. The session is designed around a guided example, your own practice, and a review of the work produced.
Questions before you join
Will the agent take live business actions?
The planned lab uses sandboxed tools. Consequential actions are reviewed by a person; production integrations are separate work.
When is the next session, and what does it cost?
Dates and fees are coming soon. The planned window is Friday, 11 a.m.–1 p.m. America/Chicago. Ask about availability below; an inquiry does not reserve a seat.
What will be confirmed before I register?
The session date, fee, capacity, delivery platform, exact software requirements, and cancellation/refund terms will be available before registration opens. Recording or replay access is not currently promised.
Can my team request a private workshop?
Yes. Choose “A private group workshop” in the inquiry form and describe your goal or group size. Elliott can discuss availability and fit.
Ask about this workshop
Interested in the next date or a session for your team? Tell Elliott what you’re looking for.