Skip to content
Method

First understand. Then prepare. Then assess.

The FT method brings together architectural judgement, Danish building data and AI-assisted preparation. Not to make the architect less important, but to create a better basis for decisions.

01

First reading

We begin with the building, place, people and intention. Data is a tool, not a substitute for the architect's first reading.

02

Source-aware data

We gather public data sources, documents and the client's own information, marking what is certain, probable or uncertain.

03

The studio knowledge layer

We connect sources, methods, experience and digital workflows so the preparation can be checked and repeated in Danish building projects.

04

Agent-assisted preparation

AI agents can prepare local development plans, BR18 requirements, building profiles and report foundations, but they must not hide uncertainty.

05

Architectural validation

Thais and the studio assess consequences, proportions, materials, use, responsibility and what should actually be recommended.

06

Clear boundaries

We distinguish between a free first reading, paid advice, homeowner dialogue and agent work for architecture studios.

Quality gate

Four questions before knowledge becomes a recommendation.

The FT method requires AI and data output to be verifiable, not merely well written. Every important finding must therefore pass through the same gate.

Source

Which public source, document or internal experience supports the information? If the source is missing, the output must say so clearly.

Status

Is the point certain, probable, uncertain or simply a question? That status must be visible before a recommendation is used with a client or studio.

Consequence

What does the information mean for the project: budget, plan, BR18, materials, everyday life, conservation, the need for advisers or the next decision?

Responsibility

Who validates the next step: architect, municipality, adviser, owner, management or the agent's technical owner? Without responsibility, AI output is too easy to misuse.

Editorial responsibility

Author
By
Published / reviewed
Published
Basis
Method and source basisPrivacy Policy: Privacy