Company profile · 4 min read
Drafted: AI home design before the architect and construction process
Drafted generates residential floor plans, 3D designs and CAD/BIM exports so people can explore custom homes before committing to a full design process.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
Drafted is an AI home-design product that turns a slow architectural starting point into something people can explore immediately. It fits homeowners, developers and design professionals who want to test possibilities before paying for a full design process.
What it does
Drafted’s current site generates floor plans, 3D models, renders and CAD/BIM exports. The user can specify rooms, square footage, footprints, lot boundaries and relationships, then iterate across designs in 2D and 3D.
The product is not trying to be a generic image generator. The useful constraint is spatial continuity: the footprint, rooms and square footage should remain coherent as the user explores options. That makes export and downstream construction more important than a pretty first render.
Why I’d look closer
The company’s story includes a real product reset. TechCrunch reports that founder Nick Donahue previously built Atmos, a more operational custom-home design business, and later started Drafted as a software-first version of the problem. That is a useful distinction: Drafted is trying to remove the service operation rather than wrap an AI model around it.
The YC launch says Drafted began with internal experience in home design and reports more than 120,000 recent users and 325,000-plus generated home designs. Those are company-reported figures. The public source also says Drafted raised $17.5 million; the investment figure is a company and press-reported fact, not evidence that the designs are buildable or that users finish projects.
The product’s public value is the exploration loop. A homeowner can get from “we want three bedrooms and a particular footprint” to several options without waiting weeks for a first sketch. A professional still needs to check codes, structure, site conditions and construction reality.
What could make it the wrong choice
The public material does not establish that a generated plan is a permit-ready construction document in every jurisdiction. It also does not make clear how lot-specific constraints, multi-story homes and local code requirements are handled across the current product.
The tradeoff is simple: Drafted may be excellent for exploring the decision before an architect, but exploration is not approval. That boundary should be explicit before a buyer treats a downloadable plan as final.
My editorial take
I would shortlist Drafted for early design exploration and for teams that want to widen the set of plausible homes before committing to a conventional process. I would not use the public usage numbers as proof that the generated plans replace architects. The interesting product is the cheap, fast first loop—not the claim that the hard parts of construction have disappeared.
Quick facts
| Field | Sourced detail |
|---|---|
| Product | AI floor plans, 3D designs and CAD/BIM exports |
| Buyers | Homeowners, developers and design workflows |
| Public claims | 120K recent users, 325K+ designs and $17.5M raised, company/press reported |
| Founder context | Prior Atmos experience described in public sources |
| Main gate | Treat generated designs as exploration until local professional review confirms otherwise |
Sources checked
| Source | Checked |
|---|---|
| YC profile | 2026-09-19 |
| Drafted homepage | 2026-09-19 |
| YC launch | 2026-09-19 |
| TechCrunch profile | 2026-09-19 |
Cohort context
Drafted is listed in Spring 2026. In our 2026-09-18 directory snapshot, 12 of 193 listed companies in that cohort have YC’s primary industry label Consumer (6.2%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:16:11.977Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Observed |
| H1 or H2 heading | Observed |
| Typed structured data | Not observed in this response |
| Docs/developer link | Not observed in this response |
| Pricing link | Not observed in this response |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗