Superbuilt AI

Knowledge Base

Everything you want
to know about the platform.

20questions answered. If yours isn't here, reach out — we respond within one business day.

ChatGPT answers questions. Superbuilt AI takes action. General LLMs have no awareness of your project files, your Revit model, your vendor list, or your local building codes. Our agents are purpose-trained on AEC workflows — they can run a clash detection pass, generate a BOQ from your drawings, draft an RFI response, and email it to the contractor. That's the gap: from conversation to execution, inside the tools you already use.

Agentic means the AI doesn't wait for you to prompt it step-by-step. You give it a goal — 'check this submittal package against the IBC 2021 and flag non-compliances' — and it breaks the task down, runs the checks across your documents, and returns a structured report. It's the difference between a chatbot and a junior consultant who actually gets things done.

No migration. Superbuilt AI connects to your existing stack — Revit, AutoCAD, Navisworks, Procore, Google Drive, Microsoft 365, email, and more. Our agents operate across these tools as a coordination layer, not a replacement. You keep your workflows; we eliminate the friction between them.

Yes. The Drawing Review Agent and Clash Detection Agent can ingest DWG, RVT, IFC, and PDF drawing sets. They parse geometry, annotations, and embedded metadata. For BIM models, the agents extract element data, run coordination checks, and flag conflicts — without requiring you to manually export or reformat anything.

The agent is trained on a live, versioned code library covering NBC, IBC, local municipal codes, and energy standards. When a code updates, the library updates — your projects are checked against the applicable version for their jurisdiction and permit stage. You can also specify the exact code edition for a project if required by the AHJ.

The Green Certification Agent maps your design decisions against the credit requirements for LEED v4.1, GRIHA, and IGBC ratings. It tracks documentation gaps, flags missing prerequisites, and generates the compliance narratives typically required for submissions. It won't stamp your drawings, but it will do 80% of the paperwork and checklist grind.

Yes — we built a dedicated Vastu Compliance Agent specifically for the South Asian market where client briefs routinely include Vastu requirements alongside NBC compliance. It checks orientation, room placement, entry directions, and zone relationships, and produces a Vastu report you can present to the client alongside your standard compliance documentation.

Upload a hand sketch, whiteboard photo, or rough diagram. The Sketch-to-CAD Agent interprets the geometry, infers wall thicknesses and relationships, and generates a clean DWG floor plan. Output is construction-intent quality for early stage — it's not replacing your CD set, but it compresses the concept-to-first-drawing cycle from days to minutes. You refine from there.

The Moodboard and Concept Design agents take your brief — programme, site context, client personality, references — and generate visual directions for client presentations. You control the creative parameters. Think of it as a junior designer who can produce twelve options overnight so you walk into the client meeting with real choices, not one direction to defend.

The Email Agent, Meeting Agent, and Slack Agent surface inside the tools your team already uses. They don't demand a new app — they read your threads, summarize action items, flag outstanding RFIs, and log decisions. The Vendor Calling Agent can even make follow-up calls to contractors and transcribe outcomes back into your project log. The coordination happens; you just stop chasing it.

Both. The Meeting Agent can join a scheduled call, take structured notes keyed to agenda items, extract action items with owners and deadlines, and post the summary to your drive or Slack channel within minutes of the call ending. For routine vendor coordination or progress calls, it can represent your position using briefing notes you prepare beforehand.

The BOQ Agent extracts quantities directly from your drawing set or BIM model — it's not estimating, it's measuring. Accuracy is bounded by the quality of your model or drawings. On a well-developed design, quantity accuracy lands within 3–5% of a manually produced BOQ. We recommend a single quantity surveyor review pass before procurement, which takes a fraction of the time a full manual takeoff would.

Your project data is encrypted at rest and in transit. It is never used to train our models. Each firm gets isolated data environments — no project data is shared across organizations. We are SOC 2 Type II compliant and follow ISO 27001-aligned security practices. For government or sensitive institutional projects, we offer private deployment options.

You own your data, always. On contract termination, we provide a full export of all project data and documents, and purge our systems within 30 days per your written instruction. There is no lock-in on your files — they're yours in their original formats.

Most firms have their first agent running useful tasks within one week. The productivity layer — email, calendar, meeting summaries — activates on day one. Compliance and drawing review agents typically take two to three weeks to calibrate to your firm's standards and project types. We assign an onboarding specialist, not a PDF guide.

This is the most honest question in the industry right now. Superbuilt AI removes the grunt work, not the learning. Junior architects still make design decisions, manage consultant relationships, and develop judgment — they just stop spending 40% of their time on formatting schedules and chasing RFI responses. Firms using AI tools report juniors moving faster to complex tasks, not fewer complex tasks.

The platform supports concurrent multi-project operation. Each active project maintains its own context — agents working on Project A have no access to Project B's data. You can set agent priority queues and resource allocation per project. There is no hard cap on simultaneous agent operations; limits are set by your subscription tier.

The ROI case is actually stronger for smaller firms. An 8-person firm running Superbuilt AI can handle the coordination load of a 20-person firm without adding headcount. The productivity layer alone — email management, meeting summaries, document organization — saves 6–10 hours per person per week. That's roughly one full-time equivalent recovered before you even touch the design automation agents.

The licensed architect or engineer of record remains legally responsible — as they always have been. Superbuilt AI is a professional tool, not a licensed professional. Every agent output includes a confidence level and flags its own uncertainty. The platform is designed to augment professional judgment, not override it. Think of it like a very thorough intern: you still stamp and sign.

Yes. The Site Intelligence Agent pulls from GIS databases, municipal zoning maps, FAR/FSI regulations, setback rules, and site-specific environmental constraints. For Indian projects, it covers DTCP, CMDA, RERA, and local development control regulations. Input a site address or coordinates and it returns a full developability summary before your client has finished the site visit.