Public websites and product surfaces. We use GitHub Pages when static publishing is enough, and Next.js when the product needs a richer application layer.
Publisher & developer
Cognitive Biases is built by MetalHatsCats Team.
Independent developer studio building websites, public data products, knowledge systems and AI-assisted tools. Cognitive Biases is one of our open public projects: useful for people first, but structured so search engines, agents and other software can understand it too.
Small independent team
We build public systems, not just pages.
Our work sits between product development, data engineering, publishing and applied AI. Sometimes the right answer is a small static site. Sometimes it is a richer application, a reusable dataset or a graph that connects the knowledge behind both.
Structured datasets, public interfaces, provenance, release files and machine-readable resources that can be reused instead of copied by hand.
Entities, relationships, retrieval paths and AI-ready information architecture for projects where isolated pages are not enough.
Agent workflows, evaluation, automation and ML experiments when they solve a real problem. We do not add AI merely because the acronym fits in the navigation.
How we build
Keep the architecture proportional to the problem.
Use the simplest architecture that can do the job well, keep important decisions inspectable, and validate the live result.
That usually means version-controlled source, clear ownership, structured data where it helps, accessible interfaces, explicit metadata and a release path that another developer can understand later.
Why the attribution is visible
Publisher identity is part of trust.
Cognitive Biases is not presented as an academic institution. The evidence claims belong to the cited research; the editorial and engineering decisions belong to this project and its maintainer. Showing the publisher makes that boundary easier to inspect.
Work with us
Have a useful web, data or AI problem?
If you have an idea, an existing site, a repository, public data or a knowledge base that is hard to use, we are open to interesting projects. Show us what exists and what should work better.