Seeing Through the Black Box: Observability & Evaluation for Agentic Workflows

GenAI & Data

We are handing over the steering wheel towards business process automation driven by agentic AI. But how do we maintain control when the "thought process" of an agent is hidden inside a black box? When an agent fails, it’s rarely a simple error message it’s a complex chain of misinterpreted goals, faulty tool use, or logic loops. How to light up the think process of agentic AI with open source and closed source?

Time & Place

Wed, March 25

12:00 - 12:45

The Ritz-Carlton Berlin

Salon Tiergarten

Classroom Seating

Max. Capacity: 32 Seats

Meet Your Intructors

Ardian Friedrich

Technical Sales Engineer, IBM

Ardian Friedrich is a technical specialist at IBM focusing on AI governance, development and deployment of AI and business process automation based on AI. He supports customer AI projects from the idea towards the production environment and ensures that AI is not only stuck in the development phase. He is specialized on planning and implementing agentic workflows and ensures that these workflows are transparent and observable on production environments.

What To Expect

Who Is This For?

AI platform leaders
ML & LLM engineers
AI product owners
Risk & governance teams
Enterprise architects

Pre-Requisites

No Prerequisits

What You'll Learn & Do?

Making agentic decision chains observable
Debugging multi-agent workflow failures
Evaluating agent behavior beyond accuracy
Open vs closed observability tooling
From black box to control panel

Agenda & Activities

Agenda for this session

20 min presentation + Audience Q&A

Registration

In order to register to our workshops you must purchase a Platinum Pass. With the Pass you are eligible to select up to 4 workshops. If you are interested in attending only one workshop you may purchase the Gold Plus Pass.