01Independent experiment
Agent workflows
Agent Buildprint (new tab)
From a vague brief to a structured workflow: phases, traceable evidence and clear handoffs.
Dominic Hückmann
Senior Software Developer & AI Architect
Build software.
Think further with AI.
I build software and put AI to practical use. Here I share independent projects, architecture decisions and lessons from hands-on work.
My starting point
Not just asking what AI can do. Finding out how to turn it into useful software.
01 / In practice
Built to learn.
Independent projects and ongoing experiments. No polished promises — concrete things to explore.
02Independent experiment
Media & agents
Media4Agents (new tab)
An independent experiment in media workflows: how do images and other media become part of an AI agent’s work?
03Independent experiment
Explore architecture
AI Architect (new tab)
A sparring partner for AI systems, focused on assumptions, trade-offs, failure modes and traceable sources.
Looking for the tools behind the work? Find skills, templates and workflows in the collection.
Explore the tools ↗02 / Share the learning
AI Native
Engineering.
From individual prompts to understandable systems. A self-paced learning path covering context, task contracts and operating AI features.
Open learning path ↗Becoming LLM-Native
- 01
Tokens & Attention
Understand context as an architectural constraint.
- 02
Context Engineering
Clear tasks, schemas and source boundaries.
- 03
Agentic Delivery
Design for evals, tool gates and fallbacks.
03 / Notes & perspectives
Notes from the workshop.
Five starting points for my work with agents and software architecture. The complete archive lives on the blog.
Your Agent Harness Needs a Release Process
A practical field note on operating agent-harness changes like product releases: start from a trace-backed failure, change one bounded component, evaluate repeated trials and private holdouts, then promote through review with a rollback path.
Coding Agents Need Hardened Harness Evals
Permissive coding-agent benchmarks hide a boring production truth: security policy changes agent behavior. Small teams should run the same task suite under nested hardening levels and separate model failures from tasks the policy made impossible.
Harness Engineering Is Systems Engineering for AI Agents
Harness engineering is the work of turning a probabilistic model into a controlled system. This field guide maps the twelve engineering surfaces around the model and shows how to evaluate and evolve them.
Your Agent Harness Needs a Behavior Map
Harness Handbook points at a practical bottleneck in agent engineering: the behavior you want to change is scattered across prompts, state managers, tool calls, policy code, and tests. Build a behavior map before editing the harness.
Your AI Agent Is Not Reflecting. It Is Defending Its First Answer
Asking one agent to reconsider its answer often produces a more confident defense of the same mistake. A bounded challenger-and-judge loop can create real alternatives, but only if disagreement, stopping, and judge bias are engineered explicitly.
04 / Keep in touch
A good question is
a good start.
An idea, a technical question or a different perspective? Let’s talk about it.