Make it a weekend of tech in Hamburg! The Social Developers Conference takes place on March 21 — exactly one day before Agentic Conf Hamburg on March 22 — both at SAE Institute Hamburg.

What is the Social Developers Conference?

The Social Developers Club is organizing their 6th conference, bringing together 300 attendees from all backgrounds and skill levels. Expect a full day of talks, workshops, a kids track with a Tinker Lab, an XR exploration area, and much more. It’s one of Hamburg’s most welcoming and diverse tech events — open to all genders, ages, and experience levels.

We’re big fans of what the Social Developers Club is doing, and some of us from the Agentic Hamburg team will be there in person — including with family!

Your Discount: 20% Off

As friends of the Agentic Hamburg community, you get 20% off your Social Developers Conference ticket with the code:

AgenticHamburg20

Grab your ticket here: Social Developers Conference on Eventbrite

The Perfect Tech Weekend

Why choose when you can do both? Attend the Social Developers Conference on Saturday, then join us at Agentic Conf Hamburg on Sunday. Two days, two conferences, one venue, one amazing community.

See you there!

We’re excited to announce Lutz Feldhege as a speaker at Agentic Conf Hamburg on March 22nd!

About Lutz

Lutz Feldhege is an AI Enabler and consultant who built his entire consulting platform — 19 Docker containers, 7 microservices, 3 databases, and a custom MCP server — using Cursor IDE. With a background in business consulting rather than software engineering, he brings a practitioner’s perspective on what happens when agentic tools meet real business requirements.

The Talk: The MCP Playbook: How I Wired Cursor to My Entire Production Stack

Your Cursor IDE knows about your current file. Maybe your project. But does it know about your running Docker containers, your database schemas, your deployment pipeline, and your seven microservices talking to each other over an internal Docker network?

Lutz’s does. He built feldhege-stack-mcp — a custom Model Context Protocol server that connects Cursor to his entire production infrastructure: a Next.js frontend, seven FastAPI/Node.js microservices, three dedicated PostgreSQL databases, a Qdrant vector store, and an Ollama instance, all running in Docker on a VPS.

In this talk, he’ll show the MCP server architecture, walk through the key tools and resources he exposed, and demonstrate how this changes the development experience. When Cursor understands your infrastructure, it stops generating code in isolation and starts making architecture-aware suggestions.

He’ll also cover what went wrong: context window limits that forced selectivity, latency issues with remote MCP connections, and the security decisions he had to make — what an AI agent should be allowed to see in production, and what he deliberately kept out.

Expect a live walkthrough of MCP tools, real terminal sessions, and Cursor making infrastructure-aware decisions in real time.

See Lutz Live

Catch this talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Tilman Dietrich as a speaker at Agentic Conf Hamburg on March 22nd!

About Tilman

Tilman Dietrich is a freelance IT consultant from Hamburg with an unusual background: he originally trained as a pilot with Lufthansa before switching to tech – trading cockpit checklists for deployment pipelines. Today he builds projects at the intersection of AI, Salesforce, and anything that can (and should) be automated.

You can follow Tilman on LinkedIn and his website dietrich.ai.

The Talk: Automating AI Research Delivery — From Papers to Spotify & YouTube

Every week, hundreds of AI research papers are published. Keeping up is hard. Tilman built a fully automated pipeline that does it for you.

The system scrapes the top 10 papers from Hugging Face’s weekly rankings, generates a ~10-minute two-host podcast episode about the #1 paper, publishes a blog post to Substack, creates YouTube Shorts, and delivers everything as a weekly email newsletter. All of it completely hands-off, without any human intervention.

Under the hood, the pipeline combines multiple AI APIs: Claude Opus for generating accessible podcast transcripts and blog posts, Gemini TTS for multi-speaker audio synthesis, and Veo 3.1 for video generation. A key design goal is accessibility – the podcast explains technical jargon, avoids reading math notation aloud, and uses analogies to make cutting-edge research approachable for a broad audience.

In this lightning talk, Tilman will walk through the architecture, the cron-driven workflow, and the practical challenges he solved along the way – from chunking long transcripts for TTS to automatically publishing on Spotify and YouTube.

See Tilman Live

Catch this lightning talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Luise Freese as a speaker at Agentic Conf Hamburg on March 22nd!

About Luise

Luise Freese is an Azure & Power Platform Architect and awarded Microsoft Most Valuable Professional (MVP). She works with organizations that want to move beyond digital theatre, focusing on AI-enabled solutions designed to last in real-world environments. She helps teams build accessible apps, robust data models, and automation that respects governance, compliance, and human effort.

A recurring theme in her work is making the invisible visible: unmeasured work, emotional load, operational friction, and the real cost of “quick wins”. Her motto “Changing the world one app at a time” stems from her extensive work with non-profit organizations.

Alongside client work, Luise writes and speaks about productivity myths, pseudo-agile practices, AI readiness, and why many digital initiatives fail long before technology becomes the problem. She is known for combining clear technical guidance with a sharp eye for organizational patterns — and for asking the uncomfortable questions others tend to avoid.

The Talk: My Agent Went Rogue: 5 Failure Modes and How to Contain Them

LLM demos are easy. Shipping agentic systems that don’t break is not.

Once an agent is allowed to call tools, trigger workflows, or modify state, small probabilistic quirks turn into real engineering problems. In this talk, Luise walks through concrete failure modes she has encountered when integrating LLMs into real systems:

• Structured output drifting just enough to crash downstream logic
• Agents selecting the wrong tool or repeating calls
• Reflection loops exploding token usage and latency
• Prompt tweaks causing silent regressions
• Context windows filling up with irrelevant history

For each failure mode, she will show code examples and the mitigation patterns that actually work: contract layers, validation and repair loops, tool scoping, iteration caps, cost guards, and lightweight evaluation harnesses.

This is a talk about what breaks when agents leave the playground — and how to design systems that survive that transition.

See Luise Live

Catch this talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Marvin Kruse as a speaker at Agentic Conf Hamburg on March 22nd!

About Marvin

Marvin Kruse is Technical Lead at OTTO, bringing together teams, ideas, and tech in the IT Retail & Logistics division since summer 2024. He is passionate about thoughtful software architecture — both building systems from scratch and evaluating what’s already there. Distributed systems are his playground, and he believes in pragmatic solutions over perfection.

The Talk: Kiro Made Me Do It: Falling Back in Love with Coding (the Hard Way)

As Technical Lead, finding time for hands-on coding can be surprisingly hard — especially when balancing day-to-day responsibilities and life outside of work. In this talk, Marvin shares his personal journey of rediscovering the joy of building software with the help of Kiro by AWS.

Kiro’s spec-driven approach felt like a refreshing alternative to “vibe coding”: more structure, clearer intent, and faster progress from idea to running software. But that structure comes at a price. Writing precise specifications turned out to be far more challenging than expected, and blindly trusting generated output quickly led to surprises.

This talk takes an honest, experience-driven look at where spec-driven AI tools truly shine, where they introduce new kinds of complexity, and why solid software engineering fundamentals are still essential — even (and especially) when AI is doing most of the typing.

See Marvin Live

Catch this talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Nele Lea Uhlemann as a speaker at Agentic Conf Hamburg on March 22nd!

About Nele Lea

Nele Lea Uhlemann is a Developer Relations Engineer at Fiberplane. Her passion is enabling collaboration among the multiple stakeholders involved in building and running software. Having switched sides from application development to infrastructure — and now to AI-native tooling — she brings a unique perspective on how developers work and introduce AI tooling to their workflows. At Fiberplane, she is part of a team rethinking how software gets built in the age of AI.

The Talk: DIY or Delegate: A Custom Learning Skill for AI Coding Agents

Nele Lea stopped learning. Not because she quit coding — she was shipping more than ever. But AI agents were writing the interesting parts, and she was just reviewing diffs she half-understood. Especially painful when picking up a new framework or library: AI agents would just figure it out for her, and she’d walk away having learned nothing.

So she built a custom Skill that forces the balance back. Before a project starts, you tell it what you want to learn. It then classifies every task: [DIY] tasks get broken down with specific learning goals and starting points, but you still do the research and write the code. [DELEGATE] tasks are implementation work outside that learning scope and get handed off to an agent. After each phase, it generates a quiz tied to your actual project code.

A demo of one custom skill — and proof that AI can make you a better developer, not just a faster one.

See Nele Lea Live

Catch this talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Jannik Streek as a speaker at Agentic Conf Hamburg on March 22nd!

About Jannik

Jannik Streek is an Open Source Software Engineer at B310 Digital GmbH. He has been contributing to numerous open source projects for many years. Since mid-2025 he has been experimenting full-time with agentic programming, understanding what can already be achieved and identifying its remaining limitations.

The Talk: Beyond the Vibes: Lessons from Using Spec Driven Development Frameworks for Agentic Coding

Spec Driven Development (SDD) promises a better approach to agentic programming in larger software projects by focusing on the what, not the how: defining requirements from which implementation can be derived. Instead of going prompt by prompt and nudging code changes, specifications clearly separate requirements from implementation and let you think in capabilities.

Historically, specs were among the lesser loved artifacts in the software lifecycle — serving mainly collaboration, documentation, and compliance purposes. Now the tables might be turning, placing specs at the center of how we build software with AI agents.

In this talk, Jannik compares different Spec Driven Development frameworks (such as OpenSpec, SpecKit, and others), shares practical lessons from working on open source projects with them, discusses the challenges of adoption, and honestly explores when the overhead of writing specs actually pays off — and when it doesn’t.

See Jannik Live

Catch this talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Luis Bezzenberger as a speaker at Agentic Conf Hamburg on March 22nd!

About Luis

Luis Bezzenberger is a Product Manager with a background in finance and open-source software. He works on AI-native tooling and agent systems, with a focus on local models, developer workflows, and practical automation for small teams.

The Talk: A Tiny Intelligent Being Lives Inside a Box Under My Desk

For the past few months, Luis has been running a small intelligent agent on his machine using OpenClaw and local models. The agent can read and update his CRM, access Gmail and Calendar APIs, coordinate tasks across tools, and message him via Signal or WhatsApp when it needs input.

This talk shows what happens when agent infrastructure, local models, and APIs meet real personal and small-business workflows. For freelancers and small teams, privacy isn’t just a security feature — it’s an enabler. When an agent runs locally and data stays under your control, you can automate workflows involving customer data, inboxes, documents, and internal knowledge without the legal and organizational friction that often blocks cloud-only solutions.

Topics covered:

• What local models can already do reliably
• Where cloud models still clearly win
• Hybrid agent patterns (local-first, cloud-augmented)
• Practical SME use cases: CRM updates, inbox triage, scheduling, and workflow automation

Instead of building another AI interface, this is about running a small software entity that can safely operate inside your personal or company environment.

See Luis Live

Catch this talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Daniel Moll as a speaker at Agentic Conf Hamburg on March 22nd!

About Daniel

Daniel Moll is a web and AI architect for digital solutions. His focus is on breaking rigid silos and creating sustainable success through creative, pragmatic processes for organizations of all sizes — from startups to enterprises. His journey: first website in 1999, then freelancer, lecturer, agency, startup, enterprise, and back to freelancing.

The Talk: “This (web) is for everyone!” – Reclaiming the Web with Agentic Engineering

Sir Tim Berners-Lee said “This is for everyone.” But what if “everyone” now means bots, AI agents, and large platforms — and no longer us as humans?

In this lightning talk, Daniel explores how we can reclaim the web with agentic engineering: through privacy by design, clear interfaces between humans and machines, and new, minimalist approaches to the web stack. Followed by a discussion on how we as engineers can contribute to making the web more open, sustainable, and human again.

See Daniel Live

Catch this talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma

We’re excited to announce Paul Stack as a speaker at Agentic Conf Hamburg on March 22nd!

About Paul

Paul Stack is Director of Product at System Initiative and a well-known voice in the infrastructure and DevOps community. He’s spoken at events around the world about his passion for continuous integration, continuous delivery, and solid operational practices – and why they should be part of what developers and system administrators do every day.

Paul believes that reliably delivering software is more important than its development. His focus areas are the DevOps and Continuous Delivery movements, and how they help the entire business and its customers.

You can follow Paul on X (@stack72).

The Talk: From Runbooks to Models: AI-Native Operations

When development velocity increases 10x through AI-assisted coding, operations becomes the bottleneck – not because teams are doing it wrong, but because the tools weren’t designed for this pace.

Paul argues that traditional Infrastructure as Code won’t work with AI agents. It’s dominated by taste and principles, with no architectural constraints for agents to follow. Teams that build software with AI have learned this the hard way: architecture leads to stability, vibes lead to chaos. The same pattern applies to operations.

What we need is automation designed for agents as the primary consumer: transparent type systems they can introspect, extensible primitives they can compose, schemas that match the domain, inline validation, and coverage across the full operational lifecycle – not just provisioning.

In this talk, you’ll see live demos of AI agents building workflows from intent, discovering production state through typed models, and composing operations through versioned data. Paul will also discuss what this means for the SRE role: the evolution from operator to architect, from reactive to proactive, from tribal knowledge to encoded knowledge.

See Paul Live

Catch this full talk and many more at Agentic Conf Hamburg on March 22, 2026 at SAE Institute Hamburg.

👉 Get your tickets on Luma