January 2027 · Proposed workshop · Date and format to be announced
Make Your AI Agent Reliable
Better answers. Fewer failures. Less unnecessary cost.
Your agent works in a demo. Now test what happens when information is missing, instructions conflict, or a connected tool stops responding.
Learn how to evaluate quality, set approval boundaries, and reduce unnecessary AI costs while maintaining useful results.
What you’ll work on
- Test everyday requests, ambiguous questions, and unexpected inputs.
- Check whether answers are supported by approved business information.
- Set human approval and escalation rules.
- Compare response quality, speed, and cost per successful task.
- Identify unnecessary model calls and opportunities for simpler automation.
- Decide what must improve before a pilot begins.
Leave with
- A practical test set covering everyday requests and difficult cases.
- A failure checklist showing what to fix first.
- Clear rules for human approval and escalation.
- A way to assess quality, response time, and operating cost together.
- A readiness checklist for your next pilot.
Hands-on challenge: Break My Agent
Test a sample agent, uncover its weaknesses, and explain how you would improve it.
Free networking hour
Compare challenges with business owners, founders, students, and engineers working on similar problems.
Bring: Your laptop and an agent idea or prototype. A sample workflow will be available.
Free for IEEE SPS members · Non-member pricing to be announced
In-person and digital participation arrangements will be announced with each event. Food availability, exact hours, venue, prerequisites, and non-member pricing will be stated before registration opens.