Fantasy Combat Manager
A fantasy strategy game. Build a company, shape its tactics and make every encounter count.
fantasycombatmanager.comEngineering · Intelligence · Accountability
Copenhagen, Denmark
Engineering
intelligence.
Earning trust.
I build software that turns complex ideas into useful products—and research what it takes to make autonomous systems accountable.
Human purpose
Direction, evidence and trust. One connected system.
01Selected work
Ideas made tangible.
From idea to live product in hours or days—not just generated code. Built with the Principles of Earned Autonomy skills suite: human direction, recorded decisions and verified releases.
The method behind the speed
TickerGlobe
A personal investment workspace. Research, holdings and market context, brought together.
tickerglobe.com
Sniffaro
Time-based dog walks that turn familiar surroundings into new possibilities.
sniffaro.comOpen engineering
Systems behind the work.
02Principles of Earned Autonomy
Autonomy is earned.
Not assumed.
My AI governance framework connects human intent, observable autonomy and a principled stopping condition. Its skills suite puts those ideas to work in the products above.
I / Intent
Understand the destination.
Human purpose governs the work. Tools and procedures serve it—not the other way around.
II / Evidence
Make judgment visible.
Decisions, verification and reversals leave a record that can be examined and challenged.
III / Trust
Earn room to act.
Autonomy follows demonstrated judgment, with consequential decisions kept under human authority.
03Research & frameworks
Beyond capability.
Toward accountability.
Independent research into hybrid human–AI cognition and the governance of autonomous systems. Published preprints, practical frameworks and open specifications.
01 / Preprint · ZenodoHuman + AI. Thinking together.
02 / Governance frameworkAgents under human authority.
03 / Open specificationMemory between sessions.
Augmented Individual Intelligence
A new mode of hybrid human–AI cognition.
Principles of Earned Autonomy
A framework for accountable autonomy in AI agents.
Agent Context Memory
Durable intent, evidence and learning across agent sessions.
Reference implementations
Accountability in practice.
Independent evidenceAn independent audit trail.
Autonomous improvementCode. Verify. Human review.
Standalone agent skillPurpose into practice.
LLM Harness Proxy
A proxy that records LLM interactions in a hash-chained ledger before releasing responses. Evidence capture sits outside the audited agent.
AI Steward
A code improvement loop connecting operator intent, harness-captured evidence, verification and human review.
Work Skill
A compact workflow for purpose-led improvement, explicit authority and an auditable record of decisions.
04Perspective
An engineer's range.
A researcher's questions.
Nils Wendelboe Holmager
Copenhagen, Denmark
My work spans frontend, backend, data and cloud infrastructure, with production experience across fintech, pharmaceutical and medical technology environments.
I work in AI Innovation at WSAudiology. My independent research asks a complementary question: not simply what an intelligent system can do, but what makes its decisions worthy of trust.
I connect those questions to practice through software people can actually use. The product, the architecture and the reasoning behind it belong to the same body of work.
Full-stack engineeringAI systemsProduct architectureAccountable autonomy
05Connect