Articles · October 23, 2026

Your project archive is training data: what AEC firms should know before licensing drawings

An architecture or engineering firm that has been practising for fifteen years holds thousands of project records: drawing sets, specifications, RFIs, submittals, site reports and condition assessments. Almost none of it is public. That is what makes it interesting to AI developers, and it is also why licensing it takes more care than most firms expect.

Real building data is scarce in public datasets

Researchers working on AI for buildings keep running into the same problem. A 2025 Stanford paper calls "the scarcity of large-scale, semantically annotated, and diverse 3D building datasets in the public domain" a primary obstacle to progress (section 1, "Background & Summary") [1]. Its answer was to generate 6.2 million synthetic residential buildings [1]. A 2026 paper in Automation in Construction also turned to synthetic data "to address the data scarcity" in reconstructing building models, noting that most existing buildings have only 2D drawings [2].

The largest well-known public CAD dataset, ABC, holds over a million models collected from Onshape's public library, and most of them are mechanical parts [3]. Public datasets like these rarely show a full project: a coordinated drawing set, the specification behind it and the RFIs that changed it.

None of this tells you what a buyer would pay for your archive, or whether one would accept it. It does explain why firms with years of real project records are hearing about data licensing.

What sits in a typical project archive

• Drawing sets across disciplines, often at several revisions;
• specifications and addenda;
• RFIs, submittals, change orders and meeting minutes;
• site reports, inspection reports and condition assessments;
• proposals, fee schedules and correspondence.

The value is in the connections between them: a question from site, the answer, and the revised drawing that followed. The legal questions sit in the same files.

Start with copyright: who owns your drawings?

In Canada, the Copyright Act protects plans and drawings as artistic works and buildings as architectural works [4]. The Royal Architectural Institute of Canada's practice guidance explains that, in general and by default, the architect or firm is the first owner of copyright in plans, drawings and specifications. Work an employee creates as part of their job generally belongs to the employer. Work from an independent contractor generally belongs to the contractor unless agreed otherwise. Ownership can change by written agreement. A client who owns a set of plans doesn't, by that alone, own the copyright in them [5].

In the United States, under the AIA's standard owner-architect form, B101–2017, and unless the parties change it, the architect and its consultants are the authors and owners of their instruments of service, including drawings and specifications, and keep their copyrights [6]. The owner gets a non-exclusive licence to use them only for the project [6]. The AIA also warns that many owner-drafted agreements try to transfer the architect's copyright to the owner [7].

Public-sector clients can be different again. Government and other public-sector clients may take ownership of project documents, or restrict how they are used, through their own contract terms. Check those contracts first.

What this means for licensing:

1. Pull the contracts, not just the files. On some projects you may have assigned copyright to the client. You can't license what you no longer own.
2. Separate your consultants' work. Structural, mechanical and electrical drawings in your archive may belong to the consultants who made them [5][6].
3. Check freelancer and subcontractor terms. Without a written agreement, their work may still be theirs [5].

Then confidentiality, security and personal information

Copyright is only part of it. Client agreements often include confidentiality clauses, and some projects, such as defence, critical infrastructure, correctional facilities or private residences, may carry security conditions or client expectations that rule out any outside use. Read them project by project.

Project files also hold personal information: homeowners' names and addresses, staff details in minutes and emails, and signatures on forms. Canada's Privacy Commissioner says organisations should limit collection to what is necessary and use anonymised or de-identified data where possible [8]. Sharing personal information with a third party can also require the consent of the people it is about, so get privacy advice before you share any. Ask any buyer how they will filter personal information, and decide what you remove before anything leaves your systems.

A short checklist before you say yes

• Which projects do you hold full rights to, under contracts you have read?
• Which files belong to consultants, freelancers or clients?
• Which clients or projects carry confidentiality or security terms?
• What personal information is in the files, and who will remove it?
• Who in the firm, such as principals, legal or IT, must approve?

Where to start

Begin with an inventory: which systems hold your records, how many years they cover, and which project types you could clear. Then get legal advice on your contracts before talking to any buyer.

Our free Data Check takes about two minutes. You enter your company's website and a few details, and it shows an estimated range of how many business records you hold, a data-readiness score, and how you match what buyers look for. You don't upload any files. It can't tell you what your data is worth and isn't an offer: any offer comes from the buyer after they review your data, and some companies are not accepted.

Frontenac Labs is an independent referrer. We do not act for any buyer. If your company licenses data to a buyer we introduce, that buyer may pay us a referral fee. You pay us nothing. The buyer alone decides whether to accept your company and on what terms; we cannot promise any price, payout or acceptance. Full disclosure

Before licensing any data, your company must confirm that it has the legal and contractual right to do so, including under privacy law, confidentiality agreements and client contracts. Frontenac Labs does not give legal advice on this; please get your own. This article is general information, not legal advice.

Sources

1. Mayer, Vesel, Zhao and Fischer (Stanford University), SYNBUILD-3D: A large, multi-modal, and semantically rich synthetic dataset of 3D building models at Level of Detail 4, arXiv:2508.21169, 2025; quote from section 1, "Background & Summary", of the full paper. https://arxiv.org/pdf/2508.21169
2. Liang, Yabuki and Fukuda, "Fully automated synthetic BIM dataset generation using a deep learning-based framework," Automation in Construction 181 (2026) 106584. https://doi.org/10.1016/j.autcon.2025.106584
3. Koch et al., ABC: A Big CAD Model Dataset For Geometric Deep Learning, arXiv:1812.06216 (CVPR 2019). https://arxiv.org/abs/1812.06216
4. Copyright Act, R.S.C. 1985, c. C-42, s. 2 (definitions of "architectural work" and "artistic work"). https://laws-lois.justice.gc.ca/eng/acts/c-42/page-1.html
5. Royal Architectural Institute of Canada, Canadian Handbook of Practice for Architects, Appendix A: Copyright and Architects (rev. October 2019). https://chop.raic.ca/appendix-a-copyright-and-architects
6. American Institute of Architects, AIA Document B101–2017, Standard Form of Agreement Between Owner and Architect, §7.2 and §7.3 (sample). https://assets.aiacontracts.com/ctrzdweb02/zdpdfs/aia-b101-2017-sample.pdf
7. American Institute of Architects, "Understanding copyright protection for architects," 1 December 2023. https://www.aia.org/resource-center/understanding-copyright-protection-architects
8. Office of the Privacy Commissioner of Canada, Principles for responsible, trustworthy and privacy-protective generative AI technologies. https://www.priv.gc.ca/en/privacy-topics/technology/artificial-intelligence/gd_principles_ai/