Autodesk Forma Building Layout Explorer:A Practical Guide for Architects

Autodesk’s experimental generative AI can turn an early massing model into multiple floor-plan options. The real value, however, is not automatic design—it is faster, better-informed questioning before a project becomes expensive to change.

Early design is full of consequential guesses. How deep can the floor plate become before daylight quality suffers? Where should the core sit? Can the target unit mix fit without creating awkward circulation? Architects normally answer these questions through repeated studies, and each study consumes time before anyone knows whether the idea is worth developing.

Building Layout Explorer is Autodesk’s attempt to compress that loop. Released as an experimental Forma Site Design feature in June 2026, it generates and evaluates floor-plan options from a massing model. Autodesk currently describes access as limited to commercial Forma Site Design users whose data is stored in the United States, so availability should be checked before building a workflow around it.

What the AI actually contributes

The feature uses architectural context—including massing, building type and structural material—to propose layouts for comparison. This is a narrower and more useful promise than “AI designs a building.” It starts with a designer-defined envelope and returns options that help the team interrogate that envelope.

That distinction matters. A generic image generator can make a plan-shaped picture, but it does not understand your project delivery chain. Building Layout Explorer sits inside Forma Site Design, where the proposal can already be studied against site context and early performance analyses. The intended benefit is continuity: massing, layout exploration and early evaluation occur before the team commits to detailed BIM work.

A practical five-step workflow

1. Start with a defensible massing model

AI cannot rescue a weak brief. Before generating layouts, establish setbacks, access, approximate height, orientation, structural logic and the project’s commercial or programmatic targets. If the envelope is unrealistic, the generated options will simply explore the wrong problem faster.

2. Define what a “good” option means

Create a short evaluation matrix before reviewing results. For multifamily work, this could include net-to-gross efficiency, unit mix, average corridor length, number of cores, façade exposure and the proportion of deep-plan units. For offices, add lease depth, core-to-perimeter distance and subdivision flexibility.

3. Generate families of options—not dozens of near-duplicates

Look for genuinely different spatial strategies: central versus split cores, single-loaded versus double-loaded circulation, courtyard versus bar configurations, or different structural assumptions. More options are useful only when they reveal meaningful trade-offs.

4. Review the plans like an architect

Check the output at two scales. At the building scale, examine circulation, egress logic, vertical stacking and façade opportunity. At the room or unit scale, inspect proportions, privacy, furniture usability and daylight potential. A strong metric can still conceal an uncomfortable plan.

5. Carry forward the decision, not the raw output

Select the principles that survived review, document why they won, and then rebuild or refine them in the appropriate schematic-design or BIM environment. The useful deliverable is a reasoned design direction with traceable assumptions—not a floor plan accepted because it arrived quickly.

Where it can save time—and where it cannot

The strongest use case is early optioneering for repeatable building types, especially when a team needs to test several plausible arrangements before a client workshop. It can also expose a mismatch between the desired program and the chosen massing early enough to make a low-cost change.

It is less convincing when a project depends on unusual typologies, complex heritage constraints, highly specific local codes or experiential qualities that resist simple metrics. The feature also does not remove the need for coordination, documentation or liability-bearing professional review. Treat its speed as extra thinking capacity, not permission to skip design development.

Build an evaluation matrix before generating

A simple scoring framework prevents the team from choosing the most visually persuasive option instead of the most useful one. Weight the criteria according to the brief: a constrained urban residential project may prioritize net-to-gross efficiency and core count, while a premium scheme may give more weight to dual-aspect units, façade exposure and generous shared space.

Use measurable indicators where possible, but keep a written design review beside the numbers. A corridor ratio cannot describe arrival quality; a daylight score cannot judge privacy; and an efficient grid may still produce repetitive or inflexible homes. The matrix should help the team ask consistent questions—not disguise design judgment as mathematics.

For a first pilot, score each option from one to five across six categories: program fit, circulation, daylight potential, structural clarity, servicing plausibility and future flexibility. Record one risk and one opportunity for every option. That small discipline creates a more valuable audit trail than saving twenty unexplained alternatives.

Hardware and access: what architects should know

Forma Site Design is cloud-based, so this workflow does not demand the same local GPU investment as real-time rendering in Lumion or heavy desktop visualization. In practice, a stable broadband connection, a current supported browser, sufficient memory for your wider multitasking workload and a well-calibrated display matter more than a top-tier graphics card. Autodesk’s live system-requirements page should remain the final reference because browser and platform support can change.

Studios should also check data residency, account eligibility and internal AI governance before uploading project information. Autodesk publishes AI transparency cards; these are worth reviewing alongside the project’s confidentiality requirements.

How it fits into a BIM and visualization pipeline

Building Layout Explorer belongs near the beginning of the pipeline. Forma is used to test the relationship between site, massing and planning logic; the selected direction then moves into a schematic or BIM workflow for geometry, assemblies, documentation and coordination. Visualization tools should enter after the team has identified which design questions an image needs to answer.

A useful sequence is: establish site and massing in Forma, compare layout principles, transfer the chosen direction into Revit or the studio’s BIM platform, validate the model, and only then build presentation views in Lumion, Enscape, D5 Render or another renderer. Early white-model images can support internal decisions, while material-rich renders should be reserved for options that have survived planning and technical review.

This separation also protects design intent. Generative layouts provide alternatives; BIM establishes a coordinated source of truth; visualization communicates the chosen direction. Asking one tool to perform all three roles usually produces attractive output with weak traceability.

Watch the official demonstration

Should your studio try it?

Yes—if your team already uses Forma Site Design, works on multifamily or other repeatable building programs, and can evaluate generated options critically. Run a controlled pilot on a completed or non-confidential project first. Compare the AI-assisted process with your usual workflow and record time saved, option diversity, redesign effort and decision quality.

If you are evaluating the wider platform, begin with Autodesk’s official Forma Site Design overview and trial information. Do not purchase solely for this experimental feature until access, data location and its fit with your project types are confirmed.

Frequently asked questions

Is Building Layout Explorer a finished Autodesk Forma feature?

No. Autodesk describes it as experimental. Availability and capability may change as the company gathers user feedback.

Does it replace Revit or schematic design?

No. It supports early layout exploration from a massing model. Detailed modeling, coordination, compliance and documentation still belong in the appropriate professional workflow.

Do I need a powerful GPU?

Not primarily for this cloud-based workflow. Connection quality, supported browser access and the demands of your other design applications are more relevant. Always verify current official system requirements.