Most architects still calculate embodied carbon by hand, using spreadsheets from 2011. ArchAI reads your building plans and generates LCA carbon reports and BREEAM assessments in minutes.
I've been on construction sites. I've sat in client meetings where the sustainability section of the presentation is a single slide with a recycling logo. Not because architects don't care — they do — but because the tools to calculate embodied carbon are either enterprise software costing €20k/year or manual spreadsheet processes that take weeks.
BREEAM assessors charge €5,000–€15,000 per project. Most small studios can't afford that until the design is locked and it's too late to change anything.
The gap was obvious: architects needed a tool that could read their actual project files and give them a fast, actionable sustainability read — early enough to change the design.
"I know roughly what my buildings emit. But I can't quantify it fast enough to include it in a pitch."
"We do the BREEAM assessment after the design is finished. By then, the structure is fixed."
I interviewed 11 architects and project managers in Portugal and the UK. I wasn't looking for feature requests — I was looking for the moments where their current workflow breaks.
For carbon calculations, the default tool is still Excel — often inherited from a senior colleague who built it in 2014.
Small studios (under 10 people) have essentially no software budget for sustainability tools outside of Revit/AutoCAD.
Time between completing a design and getting a sustainability assessment back. Often too late to act on findings.
I initially planned to build a tool where users input material quantities manually. After 5 interviews I threw that out. No one wants to re-enter what's already in their drawings. The real unlock was: what if the tool reads the PDF itself?
That's when I brought Claude in. Not as a chatbot, but as a document reader — give it a floor plan, a material schedule, or a spec doc, and extract the data needed to run a carbon analysis.
| Decision | Option considered | What I chose & why |
|---|---|---|
| Input method | Manual form with material quantities, dropdowns, m² fields | PDF upload + Claude extraction. Architects already have this data in documents — re-entering it is friction that kills adoption. |
| Output format | Raw data table, exportable CSV | Visual dashboard with LCA modules A1–D + narrative summary. Architects present to clients — they need something they can show, not just read. |
| BREEAM scoring | Full 9-category assessment upfront | Progressive disclosure — show overall score first, let users drill into categories. Reduces cognitive load on first use. |
| Tech stack | Python/FastAPI + React SPA | Next.js monorepo. API routes co-located with UI means faster iteration and simpler deployment on Vercel. |
| Pricing model | Per-report credits | Studio subscription. Architects work in teams — individual credits create friction. One invoice, whole studio. |
Upload → Analysis → Report. The entire product is this loop.
Down from an average of 2–3 weeks with traditional methods. For beta users, the first report came in under an hour — including upload, extraction, and PDF generation.
First three paying studios onboarded without any marketing. All came from direct conversations at architecture events. Retention after 90 days: 100%.
Users spent more time on the materials extraction step than on the final report. They wanted to correct Claude's readings before running the LCA — which I hadn't planned for. That became the most important feature of v2.