Hannes Hapke

Digits

BIO

As one of Digits’s principal machine learning engineers since 2020, Hannes Hapke is fully immersed in developing innovative ways to use machine learning to boost productivity for accountants and business owners.

Before joining Digits, Hannes solved machine learning infrastructure problems in various industries, including healthcare, retail, recruiting, and renewable energies. He was previously a senior machine learning scientist at Concur Labs at SAP Concur, where he explored innovative ways to improve the experience of business travelers.

Hannes is an active contributor to TensorFlow’s TFX Addons project and has co-authored multiple machine learning publications, including “Building Machine Learning Pipelines”, “Machine Learning Production Systems,” and the upcoming book on “GenAI Design Patterns”, all published by O’Reilly Media.

Hannes has been recognized as a Google Developer Expert for Machine Learning and continues to be passionate about machine learning engineering and production machine learning applications.

TITLE

The Hard Truth About AI Agents: Lessons Learned from Running Agents in Production

ABSTRACT

The recent surge of excitement around AI Agents mirrors previous waves of enthusiasm we’ve witnessed with RAGs, LLMs, and other AI technologies. While agents demonstrate remarkable capabilities in prototypes, deploying them in production presents unique challenges that are rarely discussed. In this talk, Hannes will share candid insights from Digits’ experience implementing user-facing agents in production environments.

Drawing from real-world case studies, this presentation will explore how Digits’ agents solve specific customer problems, the architectural decisions that enable reliable agent performance at scale, and the unexpected obstacles encountered along the way. Hannes will detail Digits’ agent infrastructure, including monitoring systems, and accessing internal data sources.

Attendees will leave with practical knowledge about agent orchestration, evaluation frameworks, and infrastructure design patterns that can be applied to their own production agent systems. Whether you’re considering implementing agents or looking to improve existing deployments, this talk offers hard-earned wisdom from the frontlines of production AI.