weight calc¶

Stage	Building
Status	Partial
Design status	In progress
Estimate	L
Confidence	Medium
Linear	PIN-232 ↗ PIN-408 ↗ PIN-307 ↗
Linear status	Done
Cycle	C17
Design	Design ↗
Linear epic	Drawings, Documents & Takeoff
Module	Takeoff Engine ↗

Priority: High

Scope¶

Drawing-intelligence ML layer under the Takeoff Engine: a symbol library (per-system detection templates) backs AI auto-takeoff detection; a Material Calculator turns measured PDF lengths into metal weight (LBS) via physics-derived reference tables (SMACNA duct, ASME pipe, MSS hanger spacing); and an assembly-extraction pipeline reads detail-drawing pages (e.g. a coil-trim detail) and synthesizes a reusable, project-specific component assembly rather than only dropping a generic library default. (Source: PIN-232, PIN-408, PIN-307.)

Acceptance criteria¶

Symbol library seeded per system (CD, VAV, LD-A/B, LDR-A/B, Exhaust, EF) backs the AI detection passes
Estimator selects a material spec per takeoff row; system computes straight + total LBS (with a fitting-allowance %) from SMACNA/ASME/MSS reference tables
Computed LBS snapshots onto the takeoff row and flows into the estimate line item when its unit is LBS
AI reads a detail-drawing page and extracts a structured assembly (component list: qty/size/material/spec/tag) rather than only a generic library default
Standard-detail recognition: common assemblies are recognized as variants of a canonical library entry, with project-specific overrides
Auto-fitting generation, AI-detected geometry, and auto-derived labor hours are explicitly out of scope for this pass

Conor's comments¶

This requires a lot of design discussion. This is a tough build and has to be PDF focused. We need to build training fiedls and train ai on detection and estimation.

Notes¶

Monorepo-only backend cluster (no cara-docs/design workflow); powers the Takeoff Engine - SMACNA weights, zone multipliers, symbol training, PDF revision diff