TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · Studio Practice · AGE 20—22 · FACTUAL · DURABLE
Your archive is infrastructure
The debrief you write today is the answer a system gives next year.
The idea
A finished project looks like storage: a folder closed, a drive filled. It is actually infrastructure. A studio system that answers "what did we do about west light on a narrow plot?" searches the studio's own archive — but only if the archive can be searched by what things mean, not just by what they are called. The debrief you write today is the answer a system gives next year. What is never written down cannot be found. What is filed as clutter is lost.
Why it matters
Every studio has ten years of judgment in files nobody can search. The difference between a studio with a memory and one without is classification, not software.
See it in the studio
A senior remembers that the 2019 school project solved a monsoon-drainage problem beautifully. The junior searches "drainage" and finds forty files named by sheet number. The solution exists and cannot be found. Had the debrief said "monsoon drainage, sloped site, open channel", the archive would have answered.
Watch for this
Believing that uploading the archive creates intelligence. It creates a pile. Retrieval quality depends on someone having described what each piece means — which precedent, which stage, why it was kept.
Try it
Take one finished project. Write the debrief a search system would need: problem, site condition, what was tried, what held, what failed. Then ask: could a stranger find this by design intent in a year?
Prove it
Explain how a studio archive can be searched by intent rather than by filename, and name what a debrief written today would teach a system next year.
How it works
Retrieval systems turn documents into searchable meaning and return the closest matches to a question. What comes back can only be as good as the description that went in. Whether a material enters the archive because it was explored, specified, procured, or weathered well is an architectural judgment no software makes for you. Training a model on a studio's own work is a further step, with questions of consent and ownership that the provenance strand takes up.
What this idea builds on
What this idea opens up
- Nothing yet names this as a foundation.
Sources
Open this idea on the map · The complete map · Logika · RBDS AI Lab, India · revised every edition.