This public research programme examines specific uses of AI inside architecture practice. Each report is written for business owners and project leaders, supported by evidence, and paired with a small prototype where building something provides a useful test.
The programme follows architecture work from winning projects and briefing through feasibility, design, documentation, coordination, consent, delivery and post-occupancy learning.
Each substantive claim points to sources, interviews, product tests, public data or a working prototype. Weak or incomplete evidence is identified.
Some topics need a small app, checker, dashboard or retrieval tool; others need interviews and synthesis. Each prototype tests a defined claim before any decision to develop a product.
Start with the six areas that can become practical proof fastest: site feasibility, AI governance, documentation QA, specification intelligence, practice productivity, and market pipeline intelligence. They are close enough to real pain that architecture firms can react to them, and they give me small things to build and share as I learn.
The areas below remain research directions. They are not presented as live tools in this publication.
Public demand signals can inform hiring, pitching and sector focus.
Building consent trends, construction pipeline data, tender notices, developer activity, council plan changes, and local market signals.
Firms benefit from early signals about where work is moving, tied to the regions, sectors and clients they serve.
Pipeline watch dashboard that turns public data into region, typology, and client-type signals.
Client choice changes when capital, consent or programme becomes the primary constraint.
Client interviews, RFP language, developer websites, listings, project announcements, and post-win/lost-job reviews.
AI may help a practice assess intent, risk and fit before senior staff commit time to an opportunity.
Client-brief classifier that turns a listing, email, or RFP into fit, risk, missing-info, and next-question notes.
A source-backed site pack can improve the start of concept design.
Parcel data, LiDAR, planning rules, hazards, services, listings, survey inputs, and known missing-data states.
Early design improves when constraints, uncertainty and source links are visible together.
Site-pack app: address to terrain, planning, constraints, evidence, and concept-readiness notes.
Task-level evidence shows where AI is changing architecture work.
Practice interviews, task diaries, tool testing, staff surveys, policy documents, and before/after workflow examples.
The relevant measure is whether AI changes the time, quality or risk of a defined task.
Workflow audit template that maps tasks by risk, repetition, review need, and automation potential.
Firms need clear boundaries for AI actions and professional review.
Professional conduct duties, insurer positions, firm policies, model-risk examples, and review-gate case studies.
Governance needs to fit daily project work and leave a record of use, evidence and approval.
AI use register and review-gate checklist for project teams.
Spatial interfaces give designers more direct control over AI-assisted form-making.
HCI research, mixed-initiative design tools, studio testing, sketch workflows, and observed friction in prompt-only tools.
Spatial work needs an interface that supports shaping, feedback and revision.
Sketch-house and house-to-3D experiments: draw, sculpt, generate, inspect, and revise.
AI can support detail retrieval and comparison under professional review.
Detail libraries, office standards, connection examples, reviewer interviews, common RFIs, and failed-check examples.
The strongest near-term uses are retrieval, comparison, packaging and review support.
Detail precedent finder that returns similar past details with source, assumptions, and reviewer notes.
Bounded model queries can support coordination while authoring remains in established tools.
Revit/IFC exports, model element data, BIM manager interviews, clash workflows, schedules, and Autodesk handoff paths.
AI is useful for interrogating and packaging model information within the existing model environment.
Model question layer that answers bounded questions from IFC/Revit exports with element references.
A lightweight checker can identify defined drawing omissions before issue.
Issue sheets, revision history, common coordination misses, QA checklists, and redline examples from real projects.
A second-pass checker can find omissions and inconsistencies for a person to close.
PDF drawing QA pass that checks titles, sheet references, detail callouts, revision notes, and missing legends.
Product recommendations need visible manufacturer and compliance evidence.
Product technical literature, EBOSS-style listings, manufacturer manuals, office spec masters, substitutions, and warranty documents.
Product intelligence needs to preserve the source and limits of every technical claim.
Product evidence card: source, use-case fit, constraints, warranty notes, substitutions, and unanswered questions.
AI can support evidence completeness and reduce avoidable consent RFIs.
Council checklists, RFIs, Building Code clauses, planning overlays, past consent packs, and consultant review notes.
The appropriate target is evidence completeness and RFI avoidance under professional review.
Consent evidence matrix that maps project documents to checklist requirements and missing evidence.
Visible assumptions support earlier climate and carbon decisions.
Embodied-carbon data, EPDs, material schedules, climate hazards, passive design assumptions, and rating-tool requirements.
AI can help surface trade-offs when assumptions and data quality remain visible.
Carbon and climate assumption card attached to early material or site decisions.
Early risk notes can identify avoidable procurement and buildability issues.
Quantity assumptions, supplier lead times, contractor feedback, cost plans, RFIs, substitution history, and build sequence notes.
Early risk notes are more defensible than a falsely precise concept-stage cost estimate.
Constructability memo generator for a concept option: unknowns, long-lead items, risk flags, and questions for QS/contractor.
Task-level evidence can show where time loss affects margin and quality.
Timesheets, project stage plans, fee proposals, rework causes, issue logs, and staff interviews.
AI should be assessed against project economics, delivery quality and risk.
Project-margin diagnostic that maps repeated time drains to possible automation, templates, or review changes.
Firm positioning needs to explain the value of judgement, sector knowledge and delivery.
Firm websites, project pages, pitch material, client interviews, competitor scans, and win/loss notes.
Trusted judgement, sector knowledge and delivery credibility become more important as generic production gets cheaper.
Firm positioning review that compares public proof, project evidence, sector language, and client fit.
AI training needs to reflect each role's tasks, risks and review duties.
Role interviews, software usage, common failure cases, policy gaps, training material, and project-stage responsibilities.
Directors, designers, graduates, BIM managers and project architects require different skills.
Capability matrix that maps role, task, risk level, tool pattern, and required review behaviour.
A permissioned archive can become a source-backed precedent system.
Past project files, specs, drawings, meeting minutes, lessons learned, correspondence, and office standards.
A firm's archive can become a useful AI asset when retrieval remains cited and permissioned.
Internal precedent search that answers from a small project corpus with citations and "not enough evidence" states.
Structured post-occupancy evidence can inform later design decisions.
Occupant feedback, defects, maintenance data, energy use, comfort surveys, photos, and project review notes.
AI can help structure feedback within a repeatable loop from built outcome to future design decisions.
Post-occupancy insight pack that clusters feedback into comfort, maintenance, energy, and design-decision lessons.
Tool assessment needs to include workflow fit, review, data control and handoff.
Vendor demos, pricing, product docs, user reports, workflow fit tests, integration paths, and export quality.
Model quality alone does not establish whether a tool fits review, source control, liability and handoff.
Tool scorecard comparing source traceability, review gates, export, data control, and actual practice use-case fit.
Practical scenarios can help firms prepare for AI as normal infrastructure.
Signals from current tooling, professional regulation, client expectations, delivery models, and adjacent industries.
Future scenarios are useful when they guide current decisions about capabilities, risks and investment.
Scenario workshop board for firm leaders: likely shifts, risks, bets, and capabilities to build this year.
Each report follows the same structure: thesis, relevance to architecture practice, evidence base, workflow map, risk boundary, prototype, findings and remaining questions. This keeps the programme consistent and makes the basis of each conclusion visible.