Overview
Please join us for the Building a Custom AI Agent: A Practical Walkthrough for Policy Teams webinar on Tuesday, September 22 from 2:00 – 3:00pm ET. This session is open to the public as a preview of our AI Practical Education series — FPF members receive full access to the entire series and additional benefits.
Agentic AI tools are increasingly marketed as a solution for the kind of recurring, labor-intensive research that privacy and policy teams do every week: monitoring a regulatory landscape, tracking sector developments, or producing an internal briefing. This webinar will offer a hands-on look at how to build and maintain one yourself.
In the webinar, FPF Senior Fellow, Andrew Gruen will walk through the design of a custom agentic workflow from end to end, using a custom example of a tool that ingests material from sources such as global Data Protection Authority feeds, NIST, the Federal Register, FTC announcements, arXiv, and selected trade press, then classifies and drafts entries on privacy-related AI incidents for human review. The example is illustrative, while the design choices, tradeoffs, and failure modes it surfaces are meant to be applicable to any recurring monitoring or drafting workflow.
If you bring your own Claude Code or ChatGPT Codex subscription, we’ll reserve time for you to build your own (tiny!) working agent.
What Participants Will Learn:
- How to recognize which problems are well-suited for an agentic solution
- Practical guidance on source selection: APIs, scrapers, and RSS, and when each is worth the maintenance burden
- Classification and filtering approaches that reduce false positives
- Realistic cost and hosting expectations for a small-scale production agent
- Where human editorial review remains necessary, and how to disclose LLM-generated content responsibly
- A repeatable checklist you can apply to your own monitoring or newsletter workflow
Speakers
- Andrew Gruen, Senior Fellow, Future of Privacy Forum
- Stacey Gray, Senior Director, FPF Center for Artificial Intelligence
