Skip to main content
← engineering

Bild AI Backs Blueprint Reading With $3.1M Seed to Automate Preconstruction

By Zero G Talent•

Estimating materials and hardware schedules from architectural drawings has long been one of the most tedious manual bottlenecks in commercial construction. Estimators spend days cross-referencing floor plans, schedules, and specification binders, where a single missed detail can cause thousands of dollars in overrun. San Francisco startup Bild AI is tackling that problem directly by applying computer vision and multimodal models to architectural blueprints.

Founded in 2024 and backed by Y Combinator's W25 batch, the five-person team raised a $3.1 million seed round led by Khosla Ventures, alongside Pioneer Fund, Mission Street Capital, and Ooshma Garg, as reported by Business Insider and The American Bazaar. The company argues that cutting manual friction out of preconstruction estimating can accelerate project timelines and bring down housing and commercial building costs.

Narrowing Focus to Construction Division 8

While Bild AI initially attempted to parse complete architectural plans across all building trades at once, the founders discovered that blueprint parsing demands deep trade-specific accuracy. In a presentation at DHI conNextions 2026, co-founder Roop Pal detailed how the company refined its technical scope by focusing on Division 8: openings, doors, and hardware.

Division 8 is widely considered one of the most complex subtrades in preconstruction. Architectural plan sets often span hundreds of pages where door schedules, elevations, and written specs conflict with one another. In his talk, Pal highlighted a case where Bild AI flagged a discrepancy across three separate documents: a floor plan specified a bypass door, the elevation drawing displayed a sliding door, and the hardware spec assigned hinges intended for a single-swing door. The software caught the inconsistency, prompted an automated request for information (RFI) under human supervision, and flagged the jamb requirements before ordering.

According to an integrations walkthrough, Bild AI pulls out door lists, sizes, quantities, and leaf details, while mapping hardware sets and consolidating raw line items into organized groups. Estimators can inspect doors directly located on the PDF floor plans and push the output directly into industry enterprise software, including Comsense, Avaware, SFH, Emullion, and Excel.

Commercial distributors are already deploying the tool into their daily bidding pipelines. In a customer story featuring distributor Schiller Hardware, director of estimating Aaron Smith reported that estimators who previously completed two to three project takeoffs a week increased output to seven to ten projects per week. In a side-by-side benchmark conducted by the distributor, a senior estimator required three full workdays to build a takeoff that Bild AI completed in three hours.

Builders With Computer Vision Pedigree

Bild AI was founded by two builders whose backgrounds combine practical job-site experience with high-scale computer vision engineering.

Co-founder Puneet Sukhija built his first house at age 16 and spent seven years managing residential construction across Canada, building hundreds of homes before his previous venture was acquired when he was 23. Having spent years compiling late-night takeoffs by hand, he dropped out of the computer science program at the University of British Columbia to build software tailored to physical trades.

He teamed up with Roop Pal, who graduated from Columbia University with a computer science degree at 19. Pal spent his early career as a software engineer at Google, where he served as machine learning lead on physical perimeter security systems using lasers and AI, earning ratings in the top 6% of Google engineers. He subsequently worked on traffic light detection models at Waymo and bootstrapped a logistics SaaS startup to an acquisition before co-founding Bild AI.

Engineering Roles Open in San Francisco

Bild AI operates out of an office in San Francisco with a five-person headcount. The company approaches blueprint parsing through a model-garden strategy, coordinating specialized computer vision and object detection architectures rather than relying solely on off-the-shelf generalized large language models.

To build out this architecture, the company is hiring across three technical positions on-site in San Francisco:

  • The Founding AI Engineer works on the core intelligence layer of the construction platform. The role applies computer vision, multimodal models, and LLMs to extract data from blueprints, run cost estimations, and parse permit applications, moving prototypes into production based on direct user feedback.
  • The Founding Product Engineer takes ownership of end-to-end features across a full stack of React and Python. The engineer interviews construction customers and designs user interfaces and backends that simplify complicated blueprint geometries and hardware specifications.
  • The AI/SWE Intern takes sole ownership of an engineering project, applying recent computer vision and AI models to blueprint reading and cost estimation from initial research through to production.

Both founding engineering positions offer salaries between $120,000 and $180,000 per year, while the internship offers between $3,000 and $10,000 per month.

Open Positions at Bild AI

Role Location Work Mode Salary
Founding AI Engineer San Francisco Onsite $120,000 – $180,000 / year
Founding Product Engineer San Francisco Onsite $120,000 – $180,000 / year
AI/SWE Intern San Francisco, CA, US Onsite $3,000 – $10,000 / month

See all open roles at Bild AI