disciplines
systems-design
data-infrastructure
automation
development

Building Atlas OS, a Design-to-Pro-Forma System

Client
Website
Atlas OS
Year
2026

Project Summary

Atlas OS turns a building design into a costed, trackable project. You design buildings on a parcel, and Atlas produces the cost report, vendor list, pro forma and presentation for you, then tracks the project from planning through build. It also checks every plan against the power grid using public data, so capacity gaps get solved before construction instead of after. It's human-in-the-loop by design: AI handles the paperwork in between, people handle the relationships.

01
Phase One

Geometry as the source of truth

Most project tools keep the building in one place and the numbers in another. Atlas holds the geometry itself. Buildings are modeled in Blender and rendered in Twinmotion, and every object carries one ID that reads from the network down to the site, parcel, building, level and component. That ID is derived from where the object sits in the model, never typed by hand, and it's validated on every save. Export round-trips through glTF with every property intact, so the model and the numbers can't drift apart.

Retro sci-fi illustration of a designer at a drafting table where a glowing 3D city block floats.
02
Phase Two

One pro forma, any size

The pro forma is compositional. The same system prices a single parcel or an entire multi-building campus, built from parcel records, building records and named inputs. When a cost hasn't been sourced yet, the estimate comes back unpriced instead of guessed. A number you can't trace is worse than no number.

A balance scale holding a single parcel and a whole campus, perfectly level.
03
Phase Three

Power before the permit

Atlas connects each site to its substation and feeder through public grid data and reads whether the facility actually fits the remaining headroom. When a plan exceeds capacity, the gap shows up at the planning stage, so upgrades can be planned and paid for before build time instead of the cost landing on regular residents.

Workers and a chrome android upgrading a substation before a planned site is built.
04
Phase Four

People only where people matter

Manifests and mismatch reports are managed by AI. People get a short list of only the decisions that need a human, answer them, and the system updates itself. Their day is done. Lower-level apps, like property management, attach at each level, so one system runs from the first sketch to day-to-day operations.

Androids sort paperwork while three people make a few decisions on glowing cards.
Sign Up For My Weekly Newsletter
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Back Top