Archiboost Raises $6.2 Million to Check Construction Drawings With AI

Archiboost AI, a San Francisco company that sells software to architecture, engineering and construction teams, announced on November 3, 2026 a combined $6.2 million pre-seed and seed round, according to Commercial Observer’s report. The company’s own about page lists Hanabi Capital and Marathon Venture Capital as its lead backers, with participation from RXR Arden Digital Ventures, Perceptive Ventures and Valley Capital Partners. Commercial Observer also names three angel investors: Scott Belsky, formerly chief product officer at Adobe, Aatish Nayak, formerly vice president of product at Harvey, and Augusto Marietti, chief executive of Kong AI. The company’s homepage links to a funding announcement, but that page returned a not-found error when PSV checked it on November 3, so the dollar figure rests on Commercial Observer’s report and the company’s own notice rather than a published release PSV could read.

The company says its software reviews full drawing sets for code, coordination and specification conflicts and attaches the source to each finding. Its site lists four products: drawing QA/QC, a project copilot that answers code and document questions with citations, a construction administration tool that tracks RFIs and submittals and drafts cited responses, and a library of a firm’s approved details. It integrates with Revit, Autodesk Docs, Procore, Bluebeam and Newforma, among others. The site reports more than 550,000 drawing sheets reviewed, more than 120,000 issues identified and more than $3.5 billion of project value reviewed. Those are company figures, not PSV findings. Commercial Observer reports the new money will expand the review engine and deploy a construction administration agent that pulls drawings, specifications and draw information from Autodesk Construction Cloud and Procore. The co-founders are Georgios Kontominas, chief executive and a licensed architect, and Vasu Jogani, chief technology officer.

The detail worth noticing is who is on the cap table. RXR, a New York owner and developer, invested through its venture arm, and its logo appears among Archiboost’s listed customers. Commercial Observer quotes RXR’s head of venture investing describing the tool as a second pair of eyes in RXR’s own QA/QC reviews. That makes this an owner-side story, not only a tool for architects. A coordination issue caught on a drawing costs a markup. The same issue found in the field becomes an RFI, then possibly a change order, and the owner’s contingency usually pays. Archiboost’s own pitch to general contractors is to find the would-be RFIs in the bid set.

The second detail is the testimonial from WRNS Studio, a multi-office architecture firm. On its site, Archiboost quotes WRNS saying it reviewed a 100 percent design development set on a design-build project using custom checks built from 4,000 pages of the owner’s design guidelines, specifications and reports. That is the use case an owner should think about first: the owner’s own standards, which design teams often read once, checked against every set before it goes out. PSV’s read: the tool category is moving from the architect’s desk toward the owner’s project management process, and owners who publish design guidelines now have a way to have them checked against every set, not only enforced by memory.

For a developer with projects in design, the product is an owner’s review log at each design milestone. Inputs: the drawing set at schematic design, design development and construction documents, the project specifications, the owner’s design guidelines and lessons learned from past projects, and the governing code edition for the jurisdiction. Output: one row per finding with the sheet, the conflicting reference, the owner standard or code section it relates to, the severity, the design team’s response, and the date it closed. The log lives with the owner’s project records, not only inside the architect’s tool.

The AI review, whether Archiboost or another product, produces candidate findings. The owner’s project manager triages them, the architect of record responds and decides whether each one needs a change, and the owner’s development head signs off that the set is ready for bid or permit. Before any of this, legal and IT should read the vendor’s data terms. Archiboost says it does not use private uploads to train its models. Drawings and owner guidelines can be confidential, so the owner should confirm that and settle who can access the files before uploading a set.