American startup Blueprints AI has unveiled the second generation of its system, which turns sketches and PDF documents into detailed construction drawings that take local building codes into account. The AI can simultaneously prepare documentation for architecture, structural engineering, electrical systems, ventilation, plumbing, and interior design, and then update the entire project when individual elements are changed. Final review, however, remains the responsibility of licensed professionals.

The company was founded by architect Al Dram and data analytics specialist Coco Cao. Blueprints AI was included in the Startup Battlefield 2026 list of 200 promising startups and will present its product at the TechCrunch Disrupt conference, which will take place in San Francisco from October 13 to 15.

From thesis project to AI startup

The idea for Blueprints AI first came to Dram nearly 20 years ago while he was studying architecture. In his thesis project, he explored software capable of automatically creating design documentation for school buildings. But the necessary technologies did not yet exist at the time.

The arrival of ChatGPT in 2022 prompted him to return to the idea. Six months later, Dram launched the concept for the future service, and in 2025 he officially registered the company in Santa Monica together with Cao.

By that point, the founder already had substantial hands-on experience. He earned bachelor’s and master’s degrees in architecture from the University of Tehran, studied digital design in graduate school at the European Graduate School in Switzerland, and in 2018 launched his own architectural firm, Drams Design-Build. In 2021, he expanded into the construction business as well.

It was precisely this work on real projects that helped provide the data needed to train the system. The developers used eight years’ worth of approved drawing sets accumulated by Dram’s architecture and construction company.

One project instead of dozens of separate drawings

The first version of Blueprints AI already generated architectural construction documentation. The second generation expands the service’s capabilities: a user can start with a text description or a sketch and receive a full set of drawings for several engineering and design disciplines at once.

The system covers:

· architecture and structural elements;

· mechanical systems and ventilation;

· electrical systems;

· water supply and plumbing;

· interior design.

The main change is the ability to edit the project within the system. If the user changes one element, the AI updates the related parts of the documentation and checks the changes for compliance with local building codes.

This matters because a building project is not just a set of independent images. Changing a wall, layout, or piece of engineering equipment can affect other parts of the project. Automatic updating is intended to reduce the amount of manual work and lower the likelihood of inconsistencies between drawings.

According to Dram, architects spend about 80% of their working time preparing drawings and only 20% on actual design. He hopes to shift that balance in favor of creative work.

How much design automation costs

The first version of the service costs from $3,200 to $8,500 per month, depending on the plan. Custom terms are available for large companies.

The second generation is already being used by its first customers. According to the founders, the public launch is scheduled for the week of TechCrunch Disrupt. The company is also developing a third version, which will add structural analysis and electrical panel design.

The startup is targeting two main audiences: professional architecture, engineering, and design firms, as well as developers and property owners. The service is currently offered by subscription, but the company is considering a one-time payment option for clients who need it for only a single project.

Why architects will still have to check the results

Despite the promise of creating documentation ready for permitting, Blueprints AI does not replace licensed architects. Before documents are submitted for approval, specialists must review the drawings and certify them with their professional seal.

This is a fundamental limitation: automatic checking for compliance with building codes is not equivalent to a full guarantee of a project’s safety and suitability. Building requirements vary by location, and real-world buildings require consideration of structural specifics, site conditions, and engineering risks.

For now, the developers are focusing primarily on automating repetitive operations that take up a significant share of specialists’ working time. If the system can reliably maintain consistency across the entire document set when changes are made, it will allow architects to devote more attention to layout, usability, and the quality of space.

But how well the service handles complex projects, and how often specialists have to correct its output, still remains to be assessed as it sees broader use.