Documentation

Cost Impact — AI-assisted costing

Khatam in the cost engine: the confidence indicator, manual override, intelligent takeoff from symbols and hatches, the MEP keyword hint, and the AI verbiage helper.

Cost Impact — AI-assisted costing

Confidence Indicator

Each row in the breakdown panel shows a coloured dot indicating how certain the engine is:

Key / ControlAction
Green dot — HighUser-assigned element type (Set Element Type), or IFC type with BIM dimensions
Blue dot — MediumIFC type or Revit category matched from linked 3D element; 2D quantity used
Amber dot — LowMarkup tool type fallback (no element linked or assigned)
Grey dot — Very LowGeneric fallback to MISC-001 — review manually

Manual Override

Click the pencil icon (✎) on any row in the breakdown panel to override the auto-assigned cost code with one of your own choosing. A searchable picker lists all items in the cost library. Once overridden, the row is locked (shown with a purple ✎ badge) and will not be replaced by subsequent recalculations. Use Clear override in the picker to restore automatic assignment.

Per-Element Cost in 3D View

When you select an element in the 3D viewer, the selection overlay (bottom-left HUD) shows the estimated total cost impact for that element alongside its name and category.

Intelligent Takeoff — Routing Symbols + Hatched Areas

Two purpose-built tools elevate the cost engine from a fallback-driven guess to a high-confidence assistant whenever you mark the drawing intentionally:

Key / ControlAction
Symbol stamps (Symbols toolbar)Each of the 117 symbol stamps — electrical outlets, plumbing fixtures, fire sprinklers, civil manholes, lab equipment, and more — routes to its specific CSI code at high confidence. Twelve general-annotation symbols (north arrow, section mark, grid bubble, revision tag, etc.) deliberately produce no cost row.
Material Hatch tool (Drawing toolbar, shortcut M)Polygon area that carries declared material intent — 14 materials grouped by category (walls, floors, ceilings, roofs, demolition area). Routes at high confidence. The Demolition Area material forces scope=demo regardless of any free-text scope keyword in your markup subject.
Tip: These tools route at the same high-confidence tier as a 3D-linked markup, so cost rows from symbol stamps and Material Hatch don't need a manual override. They are how an estimator working on a generic PDF (no Revit model) gets the same quality of estimate as someone working on a BIM-linked one.

3D vs Hatch Conflict Warning

If a markup is linked to a 3D element AND has a Material Hatch applied that would route to a different CSI family (e.g. an IFC wall element hatched as "Floor — Concrete"), a non-blocking toast appears explaining the disagreement. The cost row goes via the 3D linkage (higher information), but the warning lets you decide: leave it, manually override the row to the hatch's code via the pencil (✎), or remove the hatch.

MEP Keyword Hint

If any of your markups mention MEP work — pipe, conduit, duct, electrical, sprinkler, outlet, valve, VAV, FCU, boiler, chiller, and similar trade terms — and the active PDF sheet is an architectural sheet (not M-/E-/P-/FP-), a warning MEP chip appears in the Cost Impact toolbar header. MEP elements aren't drawn on architectural sheets, so cost routing on those markups will be low confidence. Click the chip to open the file picker and load the corresponding MEP discipline sheet — markups on the right sheet route at higher confidence.

Aggregate Confidence Chip

Alongside the running total in the Cost Impact toolbar header, a small brand-color dot + percentage chip shows the weighted-average confidence of all your cost rows (weighted by total cost, so a high-dollar low-confidence row drags the aggregate more than a small one). If any rows are at low or very-low confidence, a warning counter appears. Click the chip to filter the breakdown grid to only those low-confidence rows — your due-diligence list for review.

Workflow
Auditing low-confidence cost rows before sharing with a client
  1. Open the Cost Impact breakdown panel. Look at the aggregate confidence chip in the panel header (e.g. 78% · 3 low rows flagged).
    The aggregate confidence is weighted by row $-value — a single high-cost low-confidence row drags the aggregate down more than ten cheap weak-confidence rows.
  2. If the chip flags any low-confidence rows, click it once — the breakdown filters down to just the rows the engine wasn't sure about.
    Filter is local-only; closing the panel resets it. Click the chip again to clear the filter.
  3. For each flagged row that's wrong, right-click the row on canvas or in the sidebar → Set Element Type. For rows where you want to set the exact CSI code, use Override Cost Code from the right-click menu and search the cost library directly.
    Manual override locks the row against further auto-recalc. The row gets a small pencil icon next to its code so you and downstream reviewers know it was user-set.
  4. Export the report (CSV from the Cost menu, or PDF report from the toolbar). Manual overrides + their pencil indicator carry into the export so the reader can see exactly which costs are AI-resolved versus user-confirmed.

AI Verbiage Helper

When you edit a markup subject, a small sparkle icon appears next to the input. Click it to surface the top three canonical phrasings for what you've typed, drawn from a curated estimator vocabulary the cost engine's scope parser recognises cleanly.

Example: you type "fix that wall", click the sparkle, and pick a suggestion like "Repair drywall and refinish." or "Patch damaged drywall, touch-up paint." — your subject is rewritten. You can then substitute placeholder tokens ([N], [unit], [material]) with real values before pressing Enter. The helper is non-blocking — ignore it and your original text still works.

Tip: The corpus loads on first activation (~1–3 s while the on-device AI model warms up). Subsequent activations are instant. All inference runs locally on your machine — no draft text leaves your device.
Workflow
Using LocusAI to improve markup phrasing for better cost matching
  1. Click any markup in the sidebar to select it, then double-click the subject text to edit. Type a description of the work (e.g. "remove east drywall", "new copper supply line").
    The AI helper uses semantic similarity against ~50 canonical cost-engine phrases. Even rough drafts produce useful suggestions — you don't need to know the exact vocabulary.
  2. Watch for the purple LocusAI khatam icon next to the input — it starts to glow and rotate when your draft is at least 3 characters long.
    The icon spins, pauses, then spins again — that pacing is intentional. Continuous spin would read as "loading"; intermittent reads as "I have something for you."
  3. Click the khatam to open the suggestion popover. Each row carries a confidence dot in the same palette as the cost-row dots: Strong, Fair, or Weak.
    Green dot = Strong (≥50% cosine), copper dot = Fair (35-50%), amber dot = Weak (<35%). Strong matches usually fit. Fair matches are worth reviewing. Weak matches are inspiration, not action.
  4. Pick a suggestion to replace your draft. The first [placeholder] is auto-selected so you can immediately type a specific value (e.g. type drywall to replace [material], then Tab through any other slots). Press Enter to commit.
    If you draw a second markup with a description very similar to an existing one, LocusAI will surface a non-blocking dedup toast — your call whether to add as comment to the existing or keep separate.

Sample Workflows

  1. Hatching new CMU walls — pick the Material Hatch tool from the Drawing toolbar, click the material chip and select Wall — CMU, draw the polygon over the new wall area, press Enter to close. Cost row: CMU-001 at sqm × your area, high confidence.
  2. Marking a section for demolition — pick Material Hatch → Demolition Area, draw over the area. Cost row is a demo row regardless of what you type in the subject.
  3. Placing electrical fixtures — open Symbols toolbar, pick Electrical Outlet, click the desk position. Cost row: ELEC-OUTLET-001 at high confidence. Repeat for as many outlets as needed.
  4. Replacing a damaged wall section — mark the damaged area with any tool, right-click → Set Element Type → pick the right Revit category (or ensure it's 3D-linked), type subject "Replace damaged section". Cost rows: WALL-DEMOL-001 (demo) + WALL-INT-001 (install) + PAINT-001 (finish) — composite replace scope, automatic.
  5. Estimating from a free-text description — mark the area, double-click the markup subject to edit, type roughly what you mean ("fix that corner"), click the sparkle, accept a suggestion like "Repair drywall and refinish." — the cost engine now reliably parses scope=repair and produces a 40 %-of-install row.
Related topics
  • Cost Impact — OverviewLive rough-order-of-magnitude cost estimates as you mark up PDFs and 3D issues: first-time setup (calibrate scale), how it works, and the core budgeting workflows.
  • Khatam — Cost intelligenceKhatam in the cost engine: semantic CSI cost-code matching when no exact category fits, and the aggregate confidence indicator that surfaces low-confidence cost rows for a second look.