AI-Driven automation in architectural design: Reducing time, enhancing compliance

[edit] Automating sustainable choices

Sustainability in construction isn’t just about compliance; it’s about rethinking the way buildings are designed, the materials selected, and the supply chains that support them.

[edit] The challenge: repetition and regulation

Architectural design processes remain burdened by software capability (56%) and lack of knowledge of repetitive compliance tasks such as drawing checks, available tools (43%). annotation and regulatory reviews. These activities consume valuable time, limit creative focus and can inflate project costs. Despite the widespread adoption of digital tools and Building Information Modelling (BIM),inefficiencies persist in the early and technical stages of design.

This research explores whether Artificial Intelligence (AI) automation tools can help alleviate these pressures by streamlining workflows, supporting regulatory compliance and improving design quality. Three advanced AI platforms were evaluated: ChatGPT-4o, Grok and DeepSeek. Their performance was tested against drawings from the author’s Major Study Project (MSP), the design of an AI Research Centre. The study provides evidence of how AI can reduce time and cost while enhancing efficiency and offers guidance for Architectural Technology professionals considering the adoption of such tools.

[edit] The study

The study followed a three-phase mixed methods approach.

[edit] Phase 1: Literature review

A review of earlier AI tools such as Pix2PixHD, BIM2XML and GPT-4 showed they could automate elements of layout and data conversion but were restricted by limited datasets and poor scalability. More recent research, including work by Chen et al. (2024) and Chaillou (2019), highlights the growing potential of Large Language Models (LLMs), particularly when integrated with BIM and ontologies. These insights guided the decision to focus practical testing on three advanced platforms: ChatGPT-4o, Grok and DeepSeek.

[edit] Phase 2: Industry survey

A questionnaire was distributed to 23 industry professionals, primarily early career Architectural Technologists. Key findings included:

• Nearly 48% spend more than two hours each day on repetitive tasks.
• 56% identified drawing checks as the most frequently repeated activity.
• Major barriers to adopting automation included limited Architectural design processes remain burdened by software capability (56%) and lack of knowledge of repetitive compliance tasks such as drawing checks, available tools (43%).

These insights highlighted compliance checks and drawing reviews as the most promising tasks for AI automation.

[edit] Phase 3: Practical testing

The selected AI tools were applied to three technical drawings from the MSP: a Fire Strategy Plan, a Sanitaryware Layout and an Atrium Balustrade Detail. Each tool was asked to check compliance against UK Building Regulations (Approved Documents B, M and K) and relevant British Standards such as BS 8300 and BS 6180. Outputs were assessed for accuracy, regulatory relevance, clarity, speed and usability.

The practical testing revealed distinct strengths and limitations across the tools. ChatGPT-4o delivered the fastest responses and produced concise and clear summaries, which were particularly effective for early-stage validation of the Fire Strategy Plan. However, its outputs were sometimes general, with limited depth and fewer references to UK-specific regulations. Grok produced highly structured and detailed compliance feedback, identifying specific issues such as dimensional conflicts in the Sanitaryware Layout. It consistently referenced Approved Documents and British Standards, though it was slower in generating results and occasionally produced overly detailed outputs that required careful interpretation. DeepSeek excelled in user-centred and accessibility analysis, providing insights into fixture placement, signage visibility, and compliance with Part M and BS 8300. It was particularly effective when reviewing the Atrium Balustrade Detail, though it was less comprehensive for broader compliance summaries. Nonetheless, it added value in specialist areas such as ergonomics and safety.

The AI tools were directly applied to the AI Research Centre Major Study Project (MSP), with tangible impacts on design outcomes. Grok’s regulatory checks prompted refinement of escape routes and compartmentation in the fire strategy. DeepSeek highlighted accessibility concerns in the sanitary layouts, leading to revisions that improved compliance with Part M. ChatGPT-4o’s rapid summaries supported faster iteration and early validation before detailed reviews during early-stage design. This integration demonstrated that AI can save time, improve accuracy, and enhance design outcomes by addressing compliance earlier in the process.

[edit] Implications for practice

The research highlights both opportunities and cautions for the profession, offering several clear implications for practice. Tools such as ChatGPT-4o can highlight potential issues quickly, freeing time for higher-value design tasks, making it useful for early checks. The evaluation showed differing capabilities across the tools: Grok provided highly detailed, regulation-referenced compliance checks, while DeepSeek was particularly effective in reviewing accessibility and user-focused details aligned with Approved Document M and BS 8300. ChatGPT-4o proved most useful in the early design stages, offering fast, high-level summaries to guide iteration. A recurring limitation was the lack of UK-specific training within these AI tools, resulting in occasional inaccuracies and “hallucinated” outputs.

This reinforces the need for robust human review, particularly in compliance-critical areas such as fire safety and accessibility. Survey participants identified limited software capability and lack of tool knowledge as key barriers. Investment in upskilling, training, and tailored prompt strategies will be essential for practitioners to make full use of AI while ensuring regulatory accuracy

Several limitations were identified. Current AI models are not trained on UK Building Regulations, creating a risk of inaccuracies or “hallucinated” outputs. Results were also highly dependent on prompt wording, highlighting the need for adaptive prompt strategies.

Future research should explore premium AI platforms that may offer improved accuracy and better integration with BIM. There is also potential to extend AI application beyond compliance checking into areas such as site analysis, spatial planning and design optimisation. Dual-verification frameworks, combining AI with rule-based systems, could further improve trust and reliability.

The results of this study highlight the growing role of AI-driven automation in tackling inefficiencies within the architectural design process. Practical testing demonstrated that tools such as ChatGPT-4o, Grok and DeepSeek can significantly reduce the manual effort required for compliance checking and drawing reviews. Grok proved particularly strong in delivering detailed, regulation-referenced assessments, while DeepSeek offered valuable insights into accessibility and human-centred considerations, supporting compliance with Approved Document M and BS 8300. ChatGPT-4o stood out for its speed and clarity, providing concise summaries that were especially useful in the early design stages.

Despite these benefits, the study also identified clear limitations. None of the tested platforms were trained on UK-specific Building Regulations, leading to occasional inaccuracies and instances of fabricated references, highlighting the continuing need for human oversight. The reliance on standardised prompts also limited the depth of output, suggesting that more tailored prompt strategies and dual-verification frameworks would improve reliability. Future research should focus on exploring premium AI platforms with enhanced integration capabilities, as well as testing applications across a wider range of design stages such as site analysis and spatial planning.

In summary, this study confirms that AI-driven automation can substantially improve efficiency and cost-effectiveness in architectural workflows, particularly by addressing repetitive tasks such as compliance checks and drawing reviews. The application of advanced AI tools within the AI Research Centre MSP informed design refinements in areas including fire strategy and sanitary layouts, demonstrating the practical value of AI in supporting productivity, regulatory accuracy and user-focused design within Architectural Technology practice.

However, future research is needed. Could AI eventually become a standard tool in architectural compliance and design validation? How might the integration of AI evolve as regulatory frameworks and technologies advance?

This article appears in the AT Jourmnal issue 156 Winter 2025 as "AI-Driven automation in architectural design: Reducing time, enhancing compliance" and was written by Kieran Patel, Nottingham Trent University.

--CIAT

[edit] Related article on Designing Buildings

• AI building planning, design, construction and management tools list.
• Artificial intelligence.
• Artificial intelligence and civil engineering.
• Artificial Intelligence and its impact on the project profession.
• Artificial intelligence and real stupidity.
• Artificial intelligence and surveying.
• Artificial intelligence in the construction industry.
• Artificial intelligence in buildings.
• BSRIA Artificial Intelligence in Buildings white paper.
• Construction is an industry ripe for tech disruption.
• Construction project professionals and AI.
• Data Science vs Artificial Intelligence Understanding the Differences and Similarities.
• Edge AI hardware Market.
• Integration of Artificial Intelligence to Prevent Hazards in the Construction Industry.
• Mind the AI gap.
• The impact of digital on civil engineering.
• The impact of artificial intelligence on the built environment.
• Will AI ever be able to design buildings?