On-deviceAEC-tunedBundled, never billed

Khatam — the AI engine for AEC work.

The AEC-trained AI that powers LocusBIM. It finds elements and answers by meaning, sorts and links your markups, drafts the routine paperwork, and prices work as you go. Embeddable via SDK. Runs entirely on your machine, so your drawings never leave the room.

Read the Khatam docsSDK enquiries

No cloud round-tripNo usage feesWorks offline

What Khatam is

Khatam is LocusBIM's own AI, built for the way Architecture and Engineering teams actually work. It is not a general chatbot. It does the specific jobs that fill your day: understanding what a markup means, matching it to the right cost code, reading dimensions and materials out of plain notes, and finding related items by meaning instead of exact wording. Everything it surfaces is a suggestion you confirm or dismiss, never an action it takes on its own.

Khatam runs entirely on your machine. No round-trip to a cloud service, no per-use billing, no vendor lock-in. It ships with normal LocusBIM updates, and it is available via SDK for partners who want to embed it.

What Khatam does

Find anything by asking

Ask for what you want in plain language. "Select all the walls on level 2" highlights them in the model instantly, with no filter menus to dig through.

Jump to any tool

Type what you are trying to do. "Where is the calibrate tool?" takes you straight there instead of hunting through toolbars.

Answers from your own docs

Ask the app a question and get the right help section back, matched by meaning. Works fully offline.

The right stamp, ready

Open the stamp tool and the seal for that sheet's discipline is already on top. A pre-export check also flags any sheet missing a stamp, or carrying the wrong one, before the set leaves your firm.

Issues that sort themselves

New issues are categorized and color-coded by discipline as you create them, with a suggested priority and assignee.

Search the model by meaning

Searching "framing" finds "Structural Framing." Find elements by what they are, not the exact words buried in the model.

Views that name themselves

Saved views get a clear name from what is on screen, like "Level 2, Mechanical, Ducts," instead of "Saved View 7."

Sharper coordination threads

The most relevant comments rise to the top of long threads, you get a quiet heads-up before repeating a teammate, and typing @ suggests the right person for the discipline first.

Drawings and model, linked for you

Markups connect to the elements they describe. Click a cloud to fly to the element in 3D, click an element to jump back to the drawing.

Clashes and duplicates, organized

Related clashes group into one issue instead of hundreds, duplicate markups get flagged, and recurring problems across sheets are surfaced together.

Costs, estimated automatically

Every markup gets a rough cost estimate with the matching cost code, plus a confidence score so you know when to double-check.

Layers that organize themselves

One click sorts your markups into named layers by discipline, cost, and intent (Cost Items, RFIs, Demo, and more), reusing layers you already have and merging duplicates.

See what changed between revisions

Compare two versions of a drawing and get a plain-language summary of what was added, removed, and changed by meaning, like "3 doors changed material," not a pixel diff.

A style check before you share

A quick audit flags the markups missing what the rest of the set has, an absent subject, units, or a cost code, so an inconsistent markup set is easy to clean up.

Catch plan-vs-schedule mismatches

Point Khatam at a plan sheet and its schedule, and it flags where they disagree: a door drawn at one size but scheduled at another, or a tag on one sheet with no match on the other. A read-only list you review, nothing is changed.

Filter by asking

Type "pending clouds on Level 2" and the markup list narrows itself, mapping your words onto tool, status, layer, and author and searching the rest by meaning.

Markups become RFIs

Turn a markup into an RFI in one click. The title, description, and category come pre-filled from the markup text, ready for you to review before sending.

Session wrap-ups, written for you

After a live review, get a summary of the markups and comments: decisions, action items, and open questions, ready to copy or export as a Markdown file.

Scanned drawings, searchable

On-device text recognition makes scanned PDFs searchable like any other page.

Why on-device

No data leaves the machine

Project files, drawings, markups, cost data, and inference traffic stay local. Defensible for PIPEDA, CASL, and any jurisdiction with data residency requirements.

No usage fees

The model ships bundled. No API tokens to budget. No usage caps. No surprise invoices when your team scales up.

Works offline

Site visits, vault offices, jobsite trailers — Khatam works in any environment where a cloud LLM would fail.

No vendor lock-in

Not a wrapper around a third-party AI API. The model is proprietary to LocusBIM. Independence from OpenAI / Google / Anthropic pricing or policy changes.

How to recognize Khatam

Brand mark

The eight-pointed star — the khatam mark — appears wherever Khatam is ready to assist or actively working. The full canonical form on the left is the brand identity; in the product itself Khatam appears at a smaller inline size, where it communicates state through subtle animation.

Idle
Khatam is ready. Click to ask for help.
Suggestion ready
Khatam noticed something it could help with. Click to see the suggestion.
Working
Khatam is processing your request. Most inferences finish in 50–200 ms.

Khatam-driven results appear in purple toast notifications. Manual user overrides flip from purple to copper — the instant signal that the call was yours, not Khatam's. No ambiguity about which row got reviewed.

How Khatam was built

Khatam is LocusBIM's own AI, built for AEC work. The model and the features on top of it are developed in-house for the jobs A&E professionals do every day: matching markups to cost codes, sorting and linking them, drafting routine paperwork, and reading dimensions and materials out of plain notes.

Not a wrapper around a third-party AI service. Not licensed from another AI vendor. The model, the features built on it, how it is evaluated, and how it is packaged are all proprietary to LocusBIM Inc. Third-party software notices are listed at locusbim.com/legal/notices.

Why "Khatam"

Khatam (KHA-tam) is a centuries-old AEC craft: the geometric eight-pointed star recurring in Islamic architecture, Persian inlay (khatam-kari), and decorative stonework from Andalusia to the Mughal builders. A pattern of precision — eight points laid out with mathematical exactness, no ornament for ornament's sake.

We borrowed the name with intent. Architecture is precise. Estimates need to be specific. Suggestions should be confidence-scored. The mark on the page is the seal at the end of the work.

Embed Khatam in your product

The Khatam SDK is in development, a focused JavaScript surface for adding cost takeoff, AEC search by meaning, and turning plain notes into structured fields, for permitting platforms, document-management systems, and other AEC web workflows. Browser-native via ONNX Runtime Web. No backend required. Your customers' data stays in their browser.

For partner enquiries: khatam@locusbim.com

Khatam surfaces suggestions and confidence scores, not decisions. Every output — cost code, phrasing, dedup flag — is a recommendation the user is responsible for verifying before relying on it for professional engineering work. Khatam does not warrant accuracy for code compliance, contract pricing, or any other professional judgement.