AI Workflow in Architecture: Briefs, Research, and Review

An AI workflow in architecture assigns the assistant to the tasks that slow qualified designers down: reading a long brief, pulling together precedent research, and drafting written review responses. The architect keeps design judgment, code compliance review, and professional sign-off. Tools like fenbs keep the project's approved decisions and open questions in one record, so any assistant you involve reads the same starting point.

This kind of workflow does not replace design skill or licensed review. It removes repetitive preparation so thinking time goes where it matters most.

Where AI fits in an architecture practice

Architecture work divides into tasks that need professional judgment and tasks that need patient reading and organizing. AI is most useful in the second group.

• Reading a project brief and listing program requirements, open questions, and apparent conflicts.
• Organizing precedent images and case study notes into summaries by theme.
• Drafting written responses to client questions or design review comments.
• Checking a written specification against a checklist of required sections.
• Summarizing meeting notes and identifying action items.

A licensed architect must still review code compliance, sign drawings, and make design decisions. Zoning, accessibility, structural and fire-safety conclusions require professional verification, not an AI summary. The assistant is a preparation tool, not a professional of record.

Analyzing a project brief

Project briefs are often long, inconsistent, and written by people with different things in mind. An assistant can read a brief and produce a structured list: program requirements, stated priorities, open questions the client left unanswered, and apparent conflicts between sections.

Give the assistant your project brief and a clear instruction. For example: "List every space requirement, flag any room where the area is missing, and list any requirement that conflicts with another." Review the output before treating it as a design input. The assistant may miss an ambiguity that an experienced architect spots in two seconds.

Record the conclusions you accept as design inputs in your project record. If the client wants a 2,000-square-foot community room and the site allows only 1,500, write that constraint down before it gets lost in a later conversation. fenbs keeps those decisions attached to the project so an assistant in the next session reads the same agreed facts.

Organizing precedent research

Precedent research means collecting examples of buildings that solved a similar problem, then understanding how they solved it. The repetitive part is reading project descriptions and writing summary notes. The skilled part is deciding which precedents are actually relevant to your design problem.

Ask an assistant to summarize a project description, a published review, or an architect's statement. Ask it to compare two projects on a specific criterion, such as how each handles natural light in a long corridor. Check the summaries against the source before relying on them.

Keep your reference links in one place. A project record that holds the source links, the summary, and the specific lesson you drew from each precedent is more useful than a folder of images. What an AI brain is explains the broader job of keeping reusable knowledge outside a single chat.

Preparing for design review

Design reviews generate comments that need clear written responses. An assistant can draft a response matrix: each comment in one column, a proposed response in the next, and a status column for open or closed. You edit the matrix, accept or reject each response, and return it to the reviewer.

The assistant can also help prepare a presentation narrative. Give it the design rationale and ask it to produce a short client-facing summary that explains a specific decision, such as why the main entrance faces north. Review and rewrite until the words are accurate and yours.

For projects using structured approval workflows, record which review comments are open and which are closed after each meeting. An assistant that knows which comments remain open can help you track whether the design has addressed every concern before the next submission.

What AI cannot check for you

Code compliance requires professional review. An assistant can read a code section and explain what it requires, but it may misread technical language, miss local amendments, or fail to flag the interaction between two requirements. Treat code outputs as a starting point for your own check, not a conclusion.

Structural and fire-safety conclusions belong to licensed engineers. Accessibility compliance under the Americans with Disabilities Act and applicable local building codes requires a qualified review. The American Institute of Architects↗ (opens in a new tab) maintains resources on professional responsibility and the legal boundaries of practice.

Keep a rule in your project record for classes of conclusion that always need a second sign-off. Writing "structural dimensions require engineer review before drawings are issued" once means every assistant you connect reads that rule before acting.

Keeping project decisions and open questions current

Architecture projects run for months or years. Assistants change, team members change, and a decision made in month two is forgotten by month eight. Recording decisions as they are made costs almost nothing. Losing them and redoing the conversation costs a lot.

Separate decisions from open questions. "The client approved a flat roof" is a decision. "The client has not confirmed whether they want a planted roof" is an open question. Both belong in the record, but they carry different meanings for the next person who reads them.

The AI agent architecture guide goes deeper into how agents interact with project records during active work. An agent that reads the current state of decisions before acting makes fewer wrong assumptions and asks better questions.

How to do this with fenbs

1. Create a free account and add an architecture project. Write a short description of the goal and the current phase.
2. Add the key site constraints, program requirements, and confirmed client priorities as tasks. Use Decisions and rules for anything that applies throughout the project.
3. Record each approved design direction as a decision with the name of the person who approved it and the reason. Keep open questions clearly labeled as questions, not conclusions.
4. Connect your AI app over MCP from Integrations. Ask it to read the project context before each working session.
5. After each review meeting, update task statuses, close resolved comments, and save where things stand so the next session starts informed.

fenbs does not connect to CAD or BIM tools directly. It keeps the written project record: decisions, tasks, context notes, and progress. Your drawing tools stay separate.