AI slop is killing content quality.
We’re reviving it.
The Duo Framework: Where human soul meets AI efficiency.
The internet has an AI content problem
Mass-produced AI content is flooding the web. Generic articles. Fabricated details. Zero editorial oversight.
Readers can smell it. Search engines are catching on. And quality publishers are stuck choosing between two bad options: reject AI entirely and fall behind, or surrender to the slop machine.
There’s a third way.
Human-AI collaboration, done right
The Duo Framework puts humans back in creative control – not as reviewers of AI output, but as partners in every step of the process.
Direct. Discuss. Rewrite. Improve. Full creative freedom, with AI as your collaborator.
The result? Content that benefits from AI efficiency while keeping the soul that only human judgment can provide.
This isn’t about hiding AI use. It’s about being proud of using AI well.
The Duo Framework
A complete system for human-AI content collaboration, developed through hundreds of articles at Science Reader.
Duowriting
The practice of human-AI collaboration. Not “prompt and publish” – genuine creative partnership where the human directs, discusses, rewrites, and improves at every step.
The Duo Method
The quality system that ensures excellence: research protocols that separate source gathering from evaluation, editorial accountability with human decision points, verification layers including fact-checking and soul audits, and structured revision that turns drafts into polished content.
Evidence Tiers
A system preventing AI fabrication. Different claims require different evidence. No inventing vivid details. No unsourced personality traits. Every claim matched to appropriate proof.
Duodesk
The tool that implements it all. Coming soon.
The Workflow: Ten stages from intake to publication. Three human control points. Multiple quality gates. No shortcuts.
See it in action
Science Reader is a popular science publication built entirely on the Duo Framework.
Real articles. Real readers. Real proof that human-AI collaboration can produce content people want to read – even knowing AI helped create it.
What we learned:
Soul can be systematized through personas, style guides, and quality checks. Verification is non-negotiable – AI confidently fabricates without it. Human creativity throughout (not just at checkpoints) makes the difference. Transparency works when the content is genuinely good.
Who’s behind this
I’m Tormod Guldvog. 30 years in science communication – radio reporter, book reviewer, web editor, founder of early science websites.
When AI content tools emerged, most people saw generic, soulless output and concluded AI content was inherently bad.
I thought: maybe we’re just doing it wrong.
Two years and hundreds of articles later, I found approaches that work. The Duo Framework captures what I learned. Duodesk will make it accessible to others.
Based in Oslo, Norway. Working in English and Norwegian.
Ready to create content worth reading?
Want to talk?
Let’s discuss how the Duo Framework could work for you.
Want the tool?
Be first to know when Duodesk launches.
Want to learn?
Occasional thoughts on human-AI content done right.