Site, massing & feasibility AI
Before a single plan is drawn, AI can read a plot's sun, wind and noise and stack a rough building in hours — testing potential that used to take a fortnight. The FAR caveat stays yours.
A plot came in on Monday. By Wednesday they knew its sun, its wind, and roughly how much building it could hold.
A developer in Bengaluru sends a studio a raw plot and one question: is it worth buying? In the old world that's a fortnight of survey, massing studies and area sums. The studio drops the site into Autodesk Forma — sun hours, wind flow and noise read straight off the context — then runs Snaptrude from the RFP to a structured programme and a stacked, rough massing with code-aware dimensions. By Wednesday they're showing the developer three massing options and an honest read of the plot's potential. The speed is real and transformative. The FAR number on the slide, though, is still something a human must confirm against the actual development control rules — and that hasn't changed.
Read the site, stack the building, test the plot — in hours, not weeks
Sun, wind, noise and context, analysed before you draw anything
Early-stage site AI front-loads the environmental homework. Autodesk Forma (formerly Spacemaker) ingests a site and its context and runs analysis that used to need separate consultants and weeks: sun hours, wind flow, noise — so you can see, before committing a single line, which corner bakes, which is sheltered, where the quiet is. Digital Blue Foam brings early-stage generative urban and massing studies with a sustainability lens, proposing massing options across a site.
This matters because the biggest, most expensive design decisions — where the building sits, how it's oriented, how it's massed — are made earliest, when you traditionally have the least information. Reading the site with AI moves real environmental insight to the moment it can change the parti, instead of arriving as an afterthought in a consultant's report once the form is frozen.
The cheapest time to fix a building's orientation is before it exists. Site AI is most valuable precisely there.
Snaptrude turns a brief into a structured, stacked, code-aware massing
The second move is from words to a rough building. Snaptrude is a cloud concept-BIM tool that takes a text prompt or an RFP and walks it forward: site analysis -> a structured program (the schedule of areas the brief implies) -> code-based dimensions -> stacked stories — a massing you can read and adjust, that bridges to Revit when you go deeper.
So within hours you can go from 'mixed-use, this plot, this brief' to a stacked massing with a programme attached and rough dimensions — a feasibility study at sketch speed. That's a genuine shift: testing whether a plot can hold the developer's ambitions used to be days of manual area-stacking. Now it's an afternoon, and you can test three briefs against the same plot before lunch. The output is a strong, fast hypothesis about the plot — exactly the kind of divergence AI is built for.
The feasibility number is a hypothesis until a human confirms it locally
Here's the line you never cross. These tools compute areas, yields and 'code-aware' dimensions — but 'code-aware' rarely means your code. Most global tools do not encode India's National Building Code (NBC) or your municipality's specific FAR/FSI, setbacks, ground coverage, height limits, premium-FAR rules or zone-specific bye-laws. The landscape is explicit: most global generative tools do not know NBC or local bye-laws, so a human check is mandatory.
That means the feasibility output is a hypothesis about the plot's potential, not a sanctionable area statement. A developer who buys land on an AI's FAR number, unverified, is gambling. Your job at this stage is to take the fast, valuable massing and programme the tool gives you, then confirm every load-bearing number — FAR, coverage, setbacks, height — against the actual development control regulations for that exact plot and zone. The AI tests the plot's potential in hours; you make the number real before anyone signs a cheque.
Site-and-massing AI is your sharpest early-stage instrument: use Forma to read sun, wind and noise and Snaptrude to go RFP-to-massing, and you can give a developer a credible feasibility read in days instead of weeks. That speed wins commissions. But hold the line on the numbers: treat every FAR, coverage and setback as a hypothesis to verify against the local development control rules before it informs a purchase or a brief. The massing is the AI's; the feasibility certificate is yours, under your registration.
Massing AI is mostly an architect's tool, but the principle reaches your fit-out and adaptive-reuse work. When you assess an existing shell — a shop, a restaurant space, an office floor — the same discipline applies: use AI to read the space's light and orientation and to test programme fit fast, but verify the hard constraints (egress, occupancy, services, statutory clearances) against the real code, not the tool's assumption. The fast read tells you if the space _could_ work; your verification tells the client if it _will_.
This is where a solo or small studio can look much bigger than they are: a credible, fast feasibility study used to need a team and weeks, and now Forma plus Snaptrude can get you a defensible early read on your own. Use it to win the conversation with developers and landowners. Just never let the tool's FAR number leave your desk unverified — your reputation rides on it. Pair the AI massing with this platform's house-plan and feasibility tools, and confirm the codes yourself, every single time.
Autodesk Forma
Site analysis + massing/optioneering
Was Spacemaker. Reads a site's sun, wind and noise and supports rapid massing and early feasibility. World-class for environmental insight at concept; its area/feasibility figures still need verifying against local FAR and bye-laws.
Snaptrude
Concept BIM: RFP to program to massing
Cloud concept-BIM that takes a prompt or RFP to site analysis, a structured program, code-based dimensions and stacked stories, then bridges to Revit. 'Code-based' rarely means Indian NBC/FAR — confirm dimensions and yields locally.
Digital Blue Foam
Generative urban & massing + sustainability
Early-stage generative massing and urban studies with a sustainability lens. Strong for exploring massing options across a site; outputs are concept hypotheses, not sanctionable area statements.
“If a site-and-massing AI gives me a FAR and a buildable area for a plot, I can hand that feasibility number straight to a developer or buyer.”
No — the number is a hypothesis, not a feasibility certificate. Tools like Forma and Snaptrude compute 'code-aware' areas, but most do not encode India's NBC or your municipality's exact FAR/FSI, ground coverage, setbacks, height limits or premium-FAR rules. The massing and programme are fast, valuable divergence; the load-bearing numbers must be verified against the actual development control regulations for that specific plot and zone before anyone acts on them.
Workshop — a one-afternoon AI feasibility read on a real plot
You'll take a real (or sample) plot and produce the thing developers pay for: a fast, honest feasibility read — site analysis, a rough massing, and a programme — then mark the one number you must verify before anyone trusts it.
Better: Autodesk Forma and/or Snaptrude (trial/seat). Free fallback: a sun-path tool + manual area stacking in a spreadsheet. Your local FAR/setback rules to check against.
Run this feasibility brief through the tool (edit the brackets): PLOT: [area] sq m, [zone], context: [adjacent roads/buildings] BRIEF / RFP: [mixed-use: ground retail + 6 floors residential] TARGETS: developer wants max saleable area within rules STEP 1 SITE: get sun hours, wind, noise across the plot STEP 2 PROGRAM: derive the schedule of areas from the brief STEP 3 MASSING: stack stories; record gross area + rough FAR STEP 4 FLAG: highlight EVERY area/FAR/setback number to verify
- 1Read the site first. Run sun/wind/noise (or a sun-path tool). Note which corner overheats and which is sheltered — let that shape where you'd put what.
- 2Generate the programme and massing from the RFP. Record the gross built-up area and the rough FAR the tool implies.
- 3Pull your local rules: the real FAR/FSI, ground coverage, setbacks and height cap for that exact zone and plot.
- 4Compare the tool's numbers to the real ones. Mark each: matches, optimistic, or assumed-wrong. Almost always the tool's setback or FAR is a generic assumption, not your bye-law.
- 5Re-stack the massing to the verified rules and watch the saleable area change. That delta is the difference between a hypothesis and a feasibility study.
- 6Write the one-page feasibility note you'd actually send: the AI massing and programme up top, and a clear line — 'FAR and setbacks verified against [local DCR]; saleable area is X, not the tool's Y.' That honesty is the deliverable.
You’ll walk away with
A one-afternoon feasibility note — AI-generated site read, programme and massing, with every load-bearing number verified against local FAR and bye-laws — the exact artefact that wins a developer's trust without betting your reputation on an unverified figure.
A quick reality check, if you have ten minutes.
- 01Run the same plot through a site tool and ask it for the FAR it assumed. Then look up the real FAR for that zone. Note the gap — and that the tool never flagged the assumption.
- 02Take one plot and test two different briefs (all-residential vs mixed-use) against it. Notice how fast you can compare the plot's potential under each — that speed is the genuine win.
Site-and-massing AI moves real environmental insight and rough feasibility to the earliest, most decisive moment of a project — reading a plot's sun, wind and noise and stacking a code-aware massing in hours. That's a transformative divergence engine. But the FAR, setbacks and areas are hypotheses until a human confirms them against India's NBC and local development control rules. Test the plot fast; make the number real yourself.
Forma reads sun, wind and noise; Snaptrude takes an RFP to program to massing; Digital Blue Foam generates massing options. Together they test a plot's potential in hours, not weeks. But 'code-aware' rarely means Indian NBC/FAR — every load-bearing number is a hypothesis to verify locally before anyone acts on it.
Can AI do a site analysis and feasibility study for a plot?
Yes, remarkably fast at the early stage. Autodesk Forma reads a site's sun, wind and noise; Snaptrude takes an RFP to a structured programme and a stacked, code-aware massing; Digital Blue Foam generates massing options. Together they can test a plot's potential in hours instead of weeks. But the FAR, areas and setbacks are hypotheses — most tools don't encode India's NBC or your local development control rules, so a human must verify every load-bearing number before it informs a purchase.
Does AI massing software know Indian FAR and NBC rules?
Generally no. Tools like Forma and Snaptrude compute 'code-aware' or 'code-based' dimensions, but that usually reflects generic or international assumptions, not your municipality's specific FAR/FSI, ground coverage, setbacks, height limits or premium-FAR provisions. Most global generative tools do not encode India's NBC or local bye-laws. Treat the output as a fast, valuable feasibility hypothesis and confirm every regulatory number against the actual development control regulations for that exact zone and plot.
How does early-stage AI feasibility help win projects?
It lets you give a developer or landowner a credible read on a plot's potential in days rather than weeks — site analysis, a rough massing and a programme — which is often the conversation that wins the commission. A solo or small studio can suddenly produce work that used to need a team. The discipline that protects you is honesty about the numbers: present the AI massing as a strong hypothesis and verify the FAR and setbacks locally before anyone bets money on them.
_That closes the loop on generative design — plans, options, parametrics and now whole-plot feasibility. Next module, the least flashy and highest-ROI AI in a studio: language AI for specs, BOQs, reports and your own knowledge base._