Foldspace · a Floviken experiment

Foldspace

Foldspace is an experiment in data integrity, run in one specific corner: molecular biology, protein folding, and drug targets. AlphaFold has computed predicted structures for nearly every known protein. UniProt has curated the functional annotations. ChEMBL has logged decades of bioactivity data. The raw material is open, free, and current.

Type a UniProt ID. In 60 seconds, Foldspace returns a one-page clinical briefing — protein function, clinical relevance, approved drugs, structural confidence — assembled from UniProt, AlphaFold, and ChEMBL, fetched live every time. The briefing is downloadable as a PDF.

Rather than a clinician asking a chat model what it knows about a protein, this approach verifies the underlying data first and asks the model to write the clinical context around what's actually there. The model isn't the source. It's the translator.

AI synthesizes the language. The model is never asked to remember. It is told.

Foldspace isn't intended for clinical use. It's a thought experiment — one small attempt to work through what data integrity might look like as AI enters clinical workflows.