Agentic AI in Practice: What It Is, Where It Works, and What We Learned Letting It Build

Event Description: AI agents that don’t just answer questions but actually do work, planning tasks, using tools, and acting under human supervision, have moved from research demos to production systems in the past eighteen months. This session offers a practical introduction for IT professionals: what “agentic AI” actually means (and the spectrum it covers, from simple automation to multi-agent systems), why these systems are fundamentally different from traditional IT (stochastic, not deterministic, with everything that implies for testing and trust), where they fit (a five-criteria test you can apply to any function in your own shop), and how to deploy them safely (supervised actions, audit trails, automated quality gates, and a kill switch).

The second half grounds the concepts in two real case studies from Proofpoint. The first: a 24-hour sprint in which a content team brought a hard modernization problem to a working session in the morning, AI agents researched the architecture during the meeting, drafted the proposal that afternoon, wrote and adversarially reviewed their own build plan that evening, and built a working content pipeline overnight; 31 agents, under 90 minutes of wall-clock time. By the next morning, the content team (not engineers) was driving it themselves. The second: lessons from building Satori, Proofpoint’s enterprise agentic AI platform, where the hard problems turn out to be governance rather than technology: how agents earn their way into a catalog, why designing tools for agents is product design rather than API plumbing, planning for the traffic agents generate when adoption succeeds, and the ownership questions every large organization will face. In both, the honest lessons matter most: what the agents did well, where they needed human judgment, and why the division of labor between people and agents, not the automation itself, determined the outcome.

No AI background required. Attendees will leave with a working vocabulary, a framework for spotting good agentic-AI candidates in their own organizations, and a realistic sense of what deploying these systems safely requires.

Speaker Bio: Dan Rapp is the Chief AI and Data Officer at Proofpoint, where he drives the company’s AI vision and leads its Center of Excellence for AI innovation. With a career spanning AI, data science, and enterprise-scale computing, Dan focuses on turning innovative research into real-world, enterprise-grade solutions. His expertise spans AI-driven threat detection, privacy-preserving AI, and large-scale data architecture, with a particular focus on the safe and responsible deployment of agentic AI systems. Before Proofpoint, Dan held executive roles at Vivint Solar, FamilySearch, and other technology-driven organizations, building and scaling teams focused on applied AI, cloud computing, and digital transformation. A frequent speaker on AI governance and enterprise AI adoption, Dan’s recent work centers on what it takes to move AI agents from impressive demos to trustworthy production systems.