Staff who completed training
A policy only works if people know it exists and understand the examples.
Architecture firms · AI usage policy and governance
Clear AI rules for renderings, specs and client data that your studio will follow
Architects adopted AI tools faster than most firms wrote rules for them. Designers use Midjourney or Stable Diffusion for concept imagery, Enscape and other visualization tools add AI features, and project architects ask chat tools about code sections or spec language. Principals worry about client confidentiality, image ownership and whether a mistake could become a professional liability claim.
How it often works today
- Staff use personal ChatGPT or image generator accounts for project work, and nobody knows what client information has been entered.
- AI-generated concept images appear in client presentations without a clear position on originality, ownership or how they were made.
- Chat tools are asked about code requirements, and answers sometimes reach drawings or memos without being checked against the adopted code.
- Principals have not discussed AI use with their professional liability insurer, and contracts say nothing about it either way.
An AI usage policy sets plain rules: which tools are approved, what information can be entered, how AI-assisted output is checked and labeled, and who to ask. We write it with your team, set up approved tools with business data terms, and train staff so the policy is used rather than filed.
The improved workflow
Find out what is used today
A short anonymous survey and conversations with each studio show which tools staff already use and for what. People are more honest when the aim is better rules, not discipline.
Sort uses by risk
Low risk: drafting internal emails, summarizing your own meeting notes. Medium: concept imagery and spec language. Higher: code interpretation, anything in sealed documents, or uploading client drawings.
Write a short, practical policy
Two or three pages covering approved tools, what data can be entered, review requirements, image use in client work and who approves exceptions. Long policies do not get read.
Set up approved tools
Move common uses onto business accounts with data protection terms, such as Microsoft Copilot within your Microsoft 365 tenant, so the approved option is also the easy option.
Train and review
Run short sessions with real examples from your studio, then review the policy after 90 days and whenever a major tool or contract term changes.
How we deliver this: AI automation for small business or fractional CTO leadership, scoped in writing and run over 90 days. See pricing or the sample scope.
What we would measure
Share of AI use on approved accounts
Moving use from personal accounts to approved tools is the main data protection gain.
Exception requests and questions per month
Steady questions show people are engaging with the policy rather than ignoring it.
Where people stay in control
- Licensed architects remain responsible for code compliance and sealed documents. AI output is never relied on without checking the source.
- Client drawings and confidential project information are only entered into approved tools covered by business data terms.
- The policy is reviewed with your insurer or attorney where it touches professional liability, contracts or client agreements. This is not legal advice.
Example
Illustrative example: concept imagery for a client workshop
For example, a design team wants AI-generated mood images for a visioning workshop with a hotel client. Under the new policy, they use an approved tool, label the images as AI-assisted inspiration rather than design proposals, and avoid prompts naming living artists or competitor buildings. The project principal approves the deck before the workshop. This illustrative example shows that a good policy enables use with clear limits rather than banning it.
Tools typically involved
What the first 30 days look like
In the first 30 days we would survey current AI use, review your client contract templates and insurance questions, and sort uses by risk. Then we would draft a short policy for principal review and choose the approved tools to set up.
Common questions
Should we just ban AI tools until things settle?
Bans rarely hold. Staff keep using tools on personal devices, and the firm loses visibility. A better approach is approving a small set of tools with business data terms and clear rules on what can be entered. That reduces risk more than a ban most people quietly ignore.
Does the Colorado AI Act apply to architecture firms?
Colorado's AI law, SB 26-189 (effective January 1, 2027, replacing SB 24-205), focuses on automated decision-making technology used in consequential decisions, such as employment. Most studio uses, like drafting or imagery, are unlikely to be its main concern, but hiring tools could be. We would flag where it may be relevant and suggest you confirm with counsel.
Who owns AI-generated images we show clients?
The legal position on AI-generated images is still developing, and copyright protection for purely AI-generated work is limited in the US. That matters if a client expects to own design imagery. The policy should set how such images are labeled and used. Questions about specific contracts belong with your attorney.
The business case
Does it pay for itself?
Illustrative defaults for an architecture firm. Replace them with your numbers.
What this workflow costs today
a month in staff time, before any lost sales
Monthly fee + assumed tools +
Break-even: new clients a month (at first-year value)
Or: staff hours recovered and redeployed
Break-even additional revenue
Value of +1 point of win rate
First 90 days, service fee
Worth a closer lookTest the assumptions
Illustration only, not a forecast or guarantee. Recovered staff time becomes a cash saving only if spending actually declines or the time goes into paid work. Tool costs are an assumption; we replace every number here with yours before recommending a scope.
AI usage policy and governance in other industries
Accounting and CPA firmsFinancial advisory firmsLaw firmsMedical and specialty clinicsNonprofit organizationsTax preparation offices
AI usage policy and governance: full guide
Related guides
Questions to ask a fractional CTOAI consultant red flagsGetting staff to adopt new toolsMicrosoft Copilot for small business
Terms used here
Next step
Bring one expensive workflow. Leave with a clear next step.
In 30 minutes we look at one recurring workflow, what it costs you today and whether there is a credible case to improve it. If there is not, we will say so.
Book a 30-minute workflow review
Not ready to talk? Take the 3-minute check
30 minutes with Phil or Martin. No technical brief, no pitch deck. Phil or Martin replies personally, usually within one business day.