AI Explained: Lessons Learned from Building Agentic Systems

Описание: Enterprises are moving beyond the hype and are now in the trenches, building and deploying agentic systems. This transition from theory to practice is uncovering a new class of real-world challenges that can derail projects and erode trust. What are the hard-won lessons from the teams on the front lines?
Watch this AI Explained to learn:
– The most common failure points in agentic systems.
– Why traditional testing and Evals are insufficient for non-deterministic agents and how to shift to a continuous observability mindset.
– Practical strategies for diagnosing failures by tracing the full logic chain — from reasoning and planning to memory access and tool use.
AI Explained is our AMA series featuring experts on the most pressing issues facing AI teams.

00:00 Introductions
02:07 Reality Checks in Production Deployment
03:44 Handling Model Hallucinations and Accuracy
04:55 Types of Agent Systems and Frameworks
06:14 Use Cases and Business Value
07:48 Workflow Automation vs Autonomous Agents
10:21 Evaluation in Agent Systems
22:00 Evaluation and Observability in Agent Systems
33:19 Best Practices for Agent Development
38:20 Key Steps in Agent System Implementation
42:00 Building Stakeholder Trust in Agent Systems
44:57 Future of Agent Systems

✦ Get a customized demo: https://bit.ly/fiddler-demo ✦