How engineering and architecture firms can sell data to AI companies
Last updated How we research
In short
Engineering and architecture firms can license design calculations, specifications, design review comments, technical queries and drawings to AI companies. Under the AIA's standard B101 agreement, architects keep copyright in their drawings and license it to owners, a stronger starting position than many industries have. Defense-adjacent technical data, federal security rules and client confidentiality are the main reasons to hold records back.
Key takeaways
- AI buyers most want these records from engineering and architecture firms: (1) design calculations and the checks applied to them, (2) specifications and master spec sections and (3) design review comments, RFIs and technical query logs.
- No AI buyer publishes a price range for engineering and architecture firms yet, so value depends on volume, years of history, structure and how cleanly the rights can be shown.
- Size buyers look for: No engineering minimum is published. micro1 asks for 30 or more full-time staff, and Handshake AI for 20 or more staff and three years in business.
- The first rights issue to clear: AIA B101: the architect and its consultants keep copyright, and the owner gets a license for the project. This page covers 5 more.
- Sellers pay nothing to be assessed. We charge a success fee only when a deal completes, and the fee is agreed in writing before we introduce you to any buyer.
Engineering and architecture firms: what buyers look for
- Records buyers ask for
- Design calculations and the checks applied to them
- Specifications and master spec sections
- Design review comments, RFIs and technical query logs
- Drawings and BIM models where you hold the copyright
- Constructability reviews and design risk registers
- Lessons-learned reports and project closeout reviews
- Why buyers want them
- Engineering records show expert reasoning checked against codes and standards, with a trail from query to answer. micro1's brief asks for project histories, process documentation and templates, and an engineering project file holds all three.
- Size buyers look for
- No engineering minimum is published. micro1 asks for 30 or more full-time staff, and Handshake AI for 20 or more staff and three years in business.
- Rights issues to clear
- AIA B101: the architect and its consultants keep copyright, and the owner gets a license for the project
- ITAR and EAR deemed export rules for defense-adjacent technical data released to foreign persons
- Controlled unclassified information under 32 CFR part 2002, and classified work under the NISPOM at 32 CFR part 117
- Some owner agreements assign copyright instead, which takes a signed writing under 17 U.S.C. section 204(a)
- Subconsultants' work and manufacturers' objects inside models belong to them
- UK and EU: RIBA appointments and the Government Security Classifications Policy for UK projects
Why engineering and architecture firms work with us
- 1
AI buyers are licensing records from engineering and architecture firms to train models on real work. We tell you in one consultation whether yours are in demand.
- 2
Every buyer we introduce passes five checks, including company identity, intended use and security, before seeing even a redacted sample of your records.
- 3
You pay nothing unless a deal completes. Our success fee is agreed in writing before we introduce you to any buyer.
- 4
Your raw data never passes through us. You sign a license directly with your chosen buyer, after your own lawyer has reviewed it.
Engineering and architecture firms start from a better rights position than most. The AIA Contract Documents help center says that under B101-2017 the architect and its consultants are deemed the authors and owners of their instruments of service and retain their copyright. The obstacles are confidentiality and security, not ownership.
What engineering and architecture data do AI companies buy?
AI companies buy engineering and architecture data that captures expert reasoning with a check attached. A calculation package shows a design decision and its verification. A technical query log shows a question, the reasoning and the resolution.
micro1 asks for project histories, process documentation and templates, and says several years of records across multiple systems is generally a stronger fit. A practice with years of project files in a document management system matches that description.
| Record | Who usually owns it | Main issue |
|---|---|---|
| Calculations and design checks | The firm | Client and site identification |
| Specifications and master spec sections | The firm, unless built on a licensed master specification system | Third-party specification content |
| RFIs and technical queries | Shared across the project team | Other parties' contributions |
| Drawings and BIM models | The designer, licensed to the owner | Security-sensitive assets |
| Lessons-learned reports | The firm | Comments on named clients or contractors |
Who owns the copyright in drawings and models?
The designer usually owns the copyright in drawings and models. Under 17 U.S.C. section 201, copyright vests initially in the author, and the employer is treated as the author of work employees prepare within the scope of their employment. Under AIA B101-2017, the owner receives a license solely for constructing, using, maintaining, altering and adding to the project, according to the AIA Contract Documents help center.
Check your agreements anyway. Some owners require an assignment of copyright, which is valid only in a signed writing under 17 U.S.C. section 204(a), and design-build arrangements can change who holds what. Models also contain subconsultants' work and manufacturers' objects that are not yours to license.
Which engineering projects should you hold back?
Engineering projects involving defense work, federal security requirements or security-sensitive assets should be held back unless the client agrees in writing. Under the ITAR, at 22 CFR 120.50(a)(2), releasing technical data to a foreign person in the United States is a deemed export, and the EAR, at 15 CFR 734.13, treats a release of controlled technology to a foreign person in the same way.
Federal work adds more layers. Under 32 CFR 2002.4(h), controlled unclassified information includes information a contractor creates or possesses for the government that must be handled with safeguarding or dissemination controls, and the NISPOM at 32 CFR part 117 governs classified information disclosed to or developed by contractors. Exclude anything marked CUI or classified.
For other sensitive assets, ISO 19650-5, published in 2020, sets out a security-minded approach to managing sensitive information in BIM and across the life of an asset. Site photos and progress reports also show workers and members of the public, so remove faces and license plates, or leave the photos out.
How much is engineering data worth to AI buyers?
No buyer publishes an engineering or architecture range. The nearest published figure is Mercor's construction company listing of $115,000 to $1 million for firms with 50 to 2,000 staff, recorded by ddiy.co in October 2026, but that applies to contractors, not design firms.
micro1's general tiers start at $100,000 to $500,000 for a qualified partnership and rise to $3 million or more for highly unique operational data. If your practice also builds, see our page for construction companies.
If you hold UK or EU data
UK practices start from a similar position. RIBA Journal explains that under a RIBA appointment the architect keeps copyright and licenses it to the client, and section 11 of the Copyright, Designs and Patents Act 1988 gives copyright in employees' work to the employer. GOV.UK guidance for contractors says suppliers handle HMG classified information under the Government Security Classifications Policy, so exclude anything marked OFFICIAL-SENSITIVE or higher.
How do you sell engineering data through Sell My Data To AI?
We start with a consultation and a rights pre-check against your agreements. Every seller passes our eight-point vetting standard before a buyer sees anything, and every buyer is verified for intended use and onward transfer limits.
You approve every buyer and every license term. Read how to build a data manifest, or book a consultation.
Frequently asked questions
Do architects own the copyright in their drawings?
Can we license BIM models to AI companies?
Is there a published price for engineering data?
What about defense and federal projects?
Sources
- The rights of an architect's instruments of service, AIA Contract Documents, accessed November 2026
- 17 U.S. Code section 201: ownership of copyright, Legal Information Institute, Cornell Law School, accessed November 2026
- 17 U.S. Code section 204: execution of transfers of copyright ownership, Legal Information Institute, Cornell Law School, accessed November 2026
- 22 CFR 120.50: export, Legal Information Institute, Cornell Law School, accessed November 2026
- 15 CFR 734.13: export, Legal Information Institute, Cornell Law School, accessed November 2026
- 32 CFR 2002.4: definitions (controlled unclassified information), Legal Information Institute, Cornell Law School, accessed November 2026
- 32 CFR 117.1: purpose (National Industrial Security Program Operating Manual), Legal Information Institute, Cornell Law School, accessed November 2026
- ISO 19650-5:2020 Security-minded approach to information management, International Organization for Standardization, accessed November 2026
- Monetize your company's operational data, micro1, accessed November 2026
- How to sell company data to AI companies (2026 guide), ddiy.co, accessed November 2026
- Architects' copyright licences and infringements on drawings, RIBA Journal, accessed November 2026
- Copyright, Designs and Patents Act 1988, section 11: first ownership of copyright, legislation.gov.uk, accessed November 2026
- Government Security Classifications Policy, guidance 1.6: contractors and contracting authorities, GOV.UK (Cabinet Office), accessed November 2026
Book a free consultation for your firm
Tell us what you hold and when suits you. In 30 minutes we cover likely buyer interest, a realistic range and any rights issues to clear first. No records leave your systems.
Related guides
- How to build a data manifest buyers will trust
Selling data
- Can you legally sell company data for AI training?
Legal and compliance
- De-identification for AI training data
Legal and compliance
- What data do AI companies buy?
Selling data
- How much is your data worth to AI companies?
Pricing and value
- AI data licensing agreement checklist
Legal and compliance