How OFA Group’s AI “PlanAid” Is Transforming Building Code Compliance and Architectural Efficiency
Author(s): Scott Douglas Jacobsen
Publication (Outlet/Website): Vocal.Media
Publication Date (yyyy/mm/dd): 2025/10/14
Thomas Gaffney is the Chief Operations Officer of OFA Group (NASDAQ: OFAL), an architectural technology company pioneering AI-driven automation for building code compliance and design review. With a background in product operations and strategic partnerships, Gaffney leads initiatives that help architects, developers, and investors deploy greener, faster, and data-informed projects. Under his leadership, OFA launched its beta AI platform PlanAid in October 2025 following a successful IPO earlier that year. Gaffney frequently engages with media and industry leaders to discuss the convergence of architecture, artificial intelligence, and sustainable innovation in the built environment.
In this conversation with Scott Douglas Jacobsen, Gaffney discusses how the company’s AI platform PlanAid is transforming the slow, manual process of building code compliance into a faster, data-driven system. PlanAid reads architectural blueprints, cross-references them with local and national codes, and highlights areas of noncompliance for real-time correction. Gaffney explains how this improves project timelines, reduces costs, and enhances investor returns. He also addresses data governance, black-box AI risks, and the expansion of OFA’s tools, such as QuickBIM, into broader construction applications. The discussion underscores AI’s role in boosting efficiency while maintaining essential human oversight.
Scott Douglas Jacobsen: Today we’re here with Thomas Gaffney, Chief Operations Officer of OFA Group (NASDAQ: OFAL) an architectural practice developing proprietary AI to automate building-code compliance and accelerate design reviews. He argues that AI-led planning can shorten development timelines, reduce material and labour costs, and improve investor ROI modelling across real-estate infrastructure.
In May 2025, OFA raised capital through an IPO on the Nasdaq Capital Market, issuing 3,750,000 ordinary shares at $4.00 per share. In October 2025, the company announced the beta launch of “PlanAid,” an AI application for building-code compliance. Gaffney’s responsibilities span product operations and partnerships, helping architects, developers, and capital markets deploy greener, faster, and data-driven projects at scale. He frequently briefs the media on architecture, AI, and the convergence of technology and design. How does AI close the gap in code compliance and approvals?
Thomas Gaffney: The main factor here is time. At OFA, AI creates efficiencies and we believe AI-enabled tools will help people work more efficiently and effectively.
In terms of code compliance, one of the most time-consuming aspects of obtaining permits and approvals for a building design is ensuring that it meets code requirements. Our AI tool enables architects to input drawings—such as a 26-story hotel—and read the relevant codes from state and local jurisdictions, including fire codes. It then informs the architect or the code-compliance office about areas of non-compliance, areas of compliance, and areas where it’s not applicable. (OFA publicly describes this capability as part of its PlanAid initiative.)
That process is currently done manually and is very time-consuming. We aim to expedite it significantly—first, by allowing architects to identify areas of noncompliance before submission, and second, by enabling code-compliance officers to quickly pinpoint potential issues or confirm that the plans meet standards.
Jacobsen: Are there particular states where compliance codes, from state down to county, are incredibly stringent, and others where they’re more relaxed, where the program might encounter difficulties?
Gaffney: When it comes to safety, most standards are national. The most significant technical challenge we encountered was training the computer to interpret the lines in architectural drawings.
Interpreting the code itself isn’t that difficult, because large language models are quite effective at interpreting rules, which is essentially what code compliance is, and then applying those rules to a given object. In our case, that object is a blueprint drawn by an architect, in which there are lines, measurements, and distances everywhere.
=From a layperson’s perspective, it can be confusing with an overwhelming amount of information. The most challenging aspect for the computer was accurately interpreting those lines. Understanding the rules isn’t hard; understanding the geometry is.
Now that we’ve solved that, and the machine can accurately read the blueprint itself, applying the relevant code has become a much more straightforward process.
Jacobsen: Two questions from that. One common problem in large language models is a phenomenon known as hallucinations. Although the rates of hallucinations, since this was pointed out a couple of years ago, have been dropping, sometimes by a factor of 10 every several months. Is that a problem when interpreting compliance codes, where you have to have everything 100 percent correct? Additionally, about AI, despite its potential, what are the key cost drivers? What are the main cost savings from using AI?
Gaffney: I’ll go back to my original point on code compliance: efficiency. You’re still going to have an architect who’s trained, who’s been doing this for a long time, drawing and creating designs. You’ll also have someone in the building code compliance office reviewing those designs.
As for hallucinations, that’s where review becomes essential. It’s about helping to speed up the process. The way we’ve designed it, the system flags results: this code is good, this one is not, this one needs review.
You still have to use your professional skills. For instance, an attorney using a large language model to draft a document may obtain a fabricated case citation. They still have to check that. For example, if the model produces 900 items, you can leverage these as a starting point, but you still have to verify the citations. The same principle applies here. The AI lays out the relevant code compliance rules, the timelines, and the measurements.
For instance, take the distance between an office and an exit in a commercial building. The AI draws a direct line, notes the measured distance, and generates a box identifying the relevant code section. It shows the number in the schematic and the drawing key. The user can instantly cross-reference that with the actual code. It’s easy to double-check whether the computer’s interpretation is accurate.
To your second point, the cost savings come from time and speed. Code compliance reviews typically require three to four rounds of revisions. We aim to reduce that to one or two, thereby saving architects time up front. The longer it takes to get a building permit, the longer it takes to start construction, meaning investors wait longer for returns: no leases, no rent, no revenue. The faster the permit is approved, the quicker the building can be completed and begin operations.
Jacobsen: How will PlanAid integrate with existing workflows without introducing black-box risk for lenders and municipalities?
Gaffney: For integration, PlanAid will eventually allow you to upload designs directly into the platform. If there’s a code out of compliance for a specific room, you can adjust it manually within the platform, eliminating the need to switch to another program, recalibrate, and re-upload the file. You’ll be able to make real-time integrations and adjustments directly within the system. That saves a significant amount of time and eliminates redundant steps.
Jacobsen: Outside of architecture, what sectors will the platform touch — urban planning, private equity, portfolio construction, real estate?
Gaffney: That’s part of our long-term vision. We’re currently focused on the architectural vertical, as it’s our area of expertise. But we do plan to expand. We’re also developing another product called QuickBIM. It helps generate the schematics for electrical, mechanical, and structural engineering, as well as plumbing systems. Over time, we aim to scale our tool into broader construction applications, but it is currently purpose-built for architectural design and planning.
Jacobsen: What data governance ensures training sets stay compliant with local codes while preserving intellectual property and avoiding bias across jurisdictions?
Gaffney: The risk there is minimal because this tool isn’t designed for creative output or generative modelling. It’s not producing new intellectual property; instead, it involves interpreting public building codes from local jurisdictions and applying them to existing designs. It doesn’t generate original creative content, it validates compliance. So, IP or bias issues are mitigated mainly in this case.
Jacobsen: What’s the timeline for pilot accuracy and scaling up to municipal acceptance at large?
Gaffney: We’ve recently launched our beta platform and test cases with several architectural partners. Depending on the performance of these trials, we plan to scale rapidly.
Jacobsen: Any final thoughts based on our conversation today?
Gaffney: We believe that AI will be a significant driver of overall economic growth. The faster and more efficiently things are completed, the more room people have to create new value. The quicker projects move, the cheaper they become, and the more resources can be allocated elsewhere. AI will still need a human touch, but it’s going to enhance productivity across many industries, not just architecture.
Jacobsen: Thomas, thank you for your time today. It’s nice to meet you.
Gaffney: Thank you very much.
Last updated May 3, 2025. These terms govern all In Sight Publishing content—past, present, and future—and supersede any prior notices. In Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons BY‑NC‑ND 4.0; © In Sight Publishing by Scott Douglas Jacobsen 2012–Present. All trademarks, performances, databases & branding are owned by their rights holders; no use without permission. Unauthorized copying, modification, framing or public communication is prohibited. External links are not endorsed. Cookies & tracking require consent, and data processing complies with PIPEDA & GDPR; no data from children < 13 (COPPA). Content meets WCAG 2.1 AA under the Accessible Canada Act & is preserved in open archival formats with backups. Excerpts & links require full credit & hyperlink; limited quoting under fair-dealing & fair-use. All content is informational; no liability for errors or omissions: Feedback welcome, and verified errors corrected promptly. For permissions or DMCA notices, email: scott.jacobsen2025@gmail.com. Site use is governed by BC laws; content is “as‑is,” liability limited, users indemnify us; moral, performers’ & database sui generis rights reserved.
In-Sight Publishing 