Panel

Software Factories — Deep Dive #2 in Berlin

14. Juli 2026 · 16:00–20:00
Forum Digitale Technologien @ Salzufer 15/16 (Lanolinfabrik/Kanalhaus, EG), 10587 Berlin — showroom of Fraunhofer HHI / DFKIAuf Google Maps öffnen

Am 14. Juli 2026, 16:00–20:00 Uhr, veranstaltet das Agentic Organisations Collective sein zweites Deep-Dive-Treffen im Forum Digitale Technologien, dem Showroom von Fraunhofer HHI und DFKI in Berlin. Eine Contribution-only-Mini-Konferenz für CTOs, Lead Architects und Tech-Founder, die Software Factories real im Produktivbetrieb bauen – echte Praxis statt Pitches. Beitrag einreichen über den Call for Contributions.

Programm

17:0015 Min.

ArchitecH: A Harness Agnostic Software Factory

Burak Cemil Özafşar · Mutagent

A deep dive into the framework for SDLC automation — a harness-agnostic software factory built to run the whole delivery loop.

17:2015 Min.

The "Leafcutter" Approach: Scaling Decentralized Agentic Systems in Human Organizations

Jan-Henrik Hempel · Schafe vorm Fenster UG

A bio-inspired framework — the leafcutter-ant colony as a single decentralized organism — for deploying and scaling agentic AI systems inside human organizations: the foundational DNA of rules, division of labor, and the organizational requirements to establish, maintain and scale.

17:4012 Min.

The Automation of Automation? Generative AI and the Reconfiguration of Programming Labor

Ann Katzinski · WZB / Weizenbaum-Institut

An empirical study — interviews, case studies and netnographic data with developers worldwide — on how generative AI reconfigures programming work. Tools like Copilot and Cursor automate routine coding but do not replace human labor; they shift it toward tool selection, prompt engineering and quality assurance. The talk challenges full-substitution narratives and reframes it as automated automation: a layered reconfiguration of high-skill digital labor.

18:2020 Min.

A well-calibrated LLM-Judge: A basis for continuous improvement

Anupam Krishnamurthy · TestSolutions GmbH

A concrete method for aligning an LLM judge with human evaluators — curated ground-truth questions, alignment metrics, and a frozen immutable benchmark for drift tracking and model comparison — with a live RAG-chatbot demo.

Einen Beitrag beisteuern

Es werden keine weiteren Beiträge angenommen.

Als Besucher:in dabei sein