Every construction project generates a stack of engineering drawings, and buried inside each one is a material list that someone has to turn into a purchase order. Piping drawings, structural steel details, fitting schedules—each document contains dozens of line items specifying exactly what needs to be ordered, in what quantity, and in what unit of measure. For a typical job package with six drawings, that’s hundreds of individual material entries that need to be extracted, categorized, and consolidated before a single order goes out.
The person doing this work knows the pain intimately. They’re opening each drawing, scanning for the bill of materials section, manually keying pipe sizes and fitting counts into a spreadsheet, converting feet-and-inches measurements into consistent units, and then cross-referencing across all six drawings to sum up identical items. A 3-inch schedule 40 pipe that appears on four different drawings needs to become one line item with a combined footage total. A specific elbow fitting that shows up on three drawings needs its counts added together. Get any of this wrong and you either over-order (wasting budget) or under-order (delaying the job).
Most construction companies still do this entirely by hand. The drawings come in as PDFs, someone prints them out or opens them on a second monitor, and the transcription begins. It’s tedious, error-prone, and takes hours that could be spent on higher-value work. The question isn’t whether this process should be automated—it’s why it hasn’t been until now.
Lido is the best option for construction companies that need to extract material lists, quantities, and specifications from engineering drawings and vendor invoices automatically. It reads any document format — including complex multi-table drawings, handwritten annotations, and scanned blueprints — without templates or manual field mapping. Construction teams using Lido turn engineering drawings into purchase orders in seconds instead of hours.
Engineering drawings aren’t invoices. Most document automation tools are built for structured business documents with predictable layouts—invoices with vendor names at the top, line items in a table, totals at the bottom. Construction drawings follow a completely different logic. Material lists might appear in a corner table, a title block annotation, or spread across multiple sections of the same sheet. The formatting varies by engineering firm, by discipline, and sometimes by the individual drafter.
The dual-section problem makes things worse. A single piping drawing typically contains both shop materials and field materials (also called erection materials). Shop materials are what you need to fabricate components in a controlled environment—pipe lengths, fittings, flanges. Field materials are what the crew needs on-site for installation—bolts, gaskets, welding consumables. These two sections look nearly identical in format, but you only want to extract one of them for your purchase order. A standard OCR tool has no concept of this distinction. It sees text in a table and extracts all of it, leaving you to manually separate shop from field after the fact.
Unit-of-measure inconsistency creates consolidation headaches. Some items on a material list are counted in discrete units: 2 elbows, 4 flanges, 1 reducer. Others are measured in linear dimensions: 40 feet of 3-inch pipe, 12 feet 6 inches of 2-inch pipe. When you’re summing identical items across multiple drawings, you need the system to understand that “12’-6”” and “12 ft 6 in” are the same measurement and can be added to “8’-3”” from another drawing to get a total of 20 feet 9 inches. Traditional OCR extracts the text but has no understanding of what the numbers mean.
Selective extraction is the first requirement. When a drawing package lands on your desk, you need to pull only shop materials from the bill of materials section. The system has to understand the document well enough to identify which section is which and ignore field materials entirely. This isn’t a simple header-matching exercise—section labels vary across engineering firms, and sometimes the distinction is indicated by subtle formatting differences rather than explicit headers.
Cross-document consolidation is the second. Six drawings in a package might each list 3-inch schedule 40 carbon steel pipe. You don’t want six separate line items on your purchase order—you want one line item with the total footage summed across all six sheets. The same goes for fittings, flanges, and every other component. The system needs to recognize that “3” SCH 40 CS PIPE” on drawing 1 is the same item as “3” SCH40 CS PIPE” on drawing 4, despite minor formatting differences.
Dual quantity handling is the third. Your consolidated output needs to correctly distinguish between items measured in counts and items measured in lengths. Two elbows plus three elbows equals five elbows. Forty feet of pipe plus twelve feet six inches of pipe equals fifty-two feet six inches of pipe. Mixing these up—treating a length as a count or vice versa—produces nonsensical purchase orders that waste everyone’s time to correct.
Computed totals need to happen automatically. Once materials are extracted and matched, the quantities should be summed without manual intervention. If the same pipe description appears on six drawings, the output should show one row with the combined total, not six rows that someone has to manually add up in a spreadsheet.
The key difference is contextual understanding. Unlike traditional OCR that simply converts pixels to text, AI-powered document extraction can be trained to understand the structure and meaning of what it’s reading. When processing a piping drawing, the system learns to identify the shop materials section, ignore the field materials section, and extract only the relevant line items. This isn’t a rigid template match—it’s a flexible understanding that adapts as it encounters drawings from different engineering firms with different formatting conventions.
Prompt customization drives accuracy. Construction companies using Lido can configure extraction prompts that tell the system exactly what to look for and what to ignore. For a piping contractor, that means instructing the extraction to focus on shop material lists, skip erection materials, and capture both the item description and the quantity with its unit of measure. As the system processes more drawings from the same engineering firm or the same project, it gets better at recognizing the patterns specific to those documents.
Computed columns handle the math. After extraction, Lido’s computed columns automatically sum quantities for identical material descriptions across all drawings in the package. The system matches items by their description text, groups them together, and produces a single total for each unique material. This turns a six-drawing package with potentially hundreds of scattered line items into a clean, consolidated material list ready for purchase order creation.
The accuracy improves with use. One of the most significant advantages of AI-powered extraction is that it learns from corrections. When a construction company first starts processing their drawings, they might need to fix a few extraction errors—a misread quantity, a missed line item, a field material that slipped through. Each correction teaches the system what to look for next time. Companies using Lido for construction document automation report that accuracy improves noticeably within the first few batches of drawings, and continues improving as the system encounters more document variations.
The process starts when a drawing package arrives. A typical package contains six piping drawings for a fabrication job. Each drawing includes a bill of materials with shop materials listed separately from field materials. Before automation, an experienced estimator or project coordinator would spend hours manually transcribing each material list into a master spreadsheet, then reconciling duplicates and summing quantities. With document automation, the entire package gets uploaded at once.
Extraction happens in minutes, not hours. The system processes all six drawings, identifies the shop materials section on each one, and extracts every line item with its description, quantity, and unit of measure. Items measured in counts (fittings, flanges, reducers) are captured as discrete numbers. Items measured in lengths (pipe runs) are captured with their footage, including feet-and-inches conversions where necessary.
Consolidation is automatic. Once all six drawings are extracted, the system matches identical material descriptions and sums their quantities. A 3-inch elbow that appears on four drawings with quantities of 2, 3, 1, and 4 becomes a single line item: 3-inch elbow, quantity 10. A pipe size that appears on all six drawings with varying footages gets summed into one total length. The output is a clean, consolidated bill of materials with one row per unique item and the total quantity across all drawings.
The consolidated list feeds directly into procurement. Instead of a messy spreadsheet that needs manual cleanup, the construction company gets a purchase-order-ready material list. They can review it, make any final adjustments, and send it to their vendors. What used to take an entire morning now takes minutes of processing time plus a quick review pass.
Volume is the multiplier. A company handling one or two drawing packages a week can get by with manual transcription. It’s tedious but manageable. But construction companies don’t stay small. As project volume grows, the material takeoff bottleneck grows with it. Ten packages a week means ten mornings spent on manual data entry. Twenty packages means you’re hiring someone whose entire job is transcribing material lists from drawings into spreadsheets.
Errors compound at scale. A 2% error rate on material quantities might seem acceptable on a single drawing. But across hundreds of drawings and thousands of line items per month, that 2% translates into tens of thousands of dollars in over-orders, emergency re-orders, and project delays. Every misread quantity ripples through the supply chain—wrong amounts on purchase orders, incorrect inventory counts, fabrication delays when a critical fitting is short.
The people doing this work have better things to do. The person transcribing material lists from drawings is almost never a junior clerk—they’re usually an experienced project coordinator or estimator who understands piping specifications, material grades, and scheduling requirements. Their expertise is wasted on data entry. Automating the extraction and consolidation frees them to focus on reviewing quantities, catching engineering errors, and coordinating with vendors on lead times and substitutions.
Consistency across projects becomes possible. When material takeoff is manual, every person does it slightly differently. One coordinator might format pipe descriptions one way while another uses a different convention. This makes it harder to track material costs across projects, compare vendor pricing, or build reliable historical data. Automated extraction produces consistent output regardless of who uploads the drawings, making downstream analysis and reporting significantly more reliable.
Start with your most common document type. For piping contractors, that’s usually piping isometric drawings or plan views with material lists. For structural steel companies, it might be shop drawings with cut lists. Pick the document type you process most frequently and that causes the most manual work. This gives you the fastest payoff and the most data for the system to learn from.
Define your extraction rules clearly. The more specific you are about what you want extracted and what you want ignored, the better the results from day one. If you need shop materials only, say so explicitly. If quantities should be captured in specific units, define those units. If certain drawing sections should be skipped entirely, identify them. These rules become the extraction prompts that guide the AI.
Plan for the consolidation step. Extracting material lists from individual drawings is only half the problem. The real value comes from consolidating across multiple drawings into a single, de-duplicated list with summed quantities. Make sure your workflow includes this consolidation step, and define how you want identical items to be matched—by exact description match, by material specification, or by some combination of attributes.
Expect improvement over time. The first batch of drawings you process will require the most review and correction. That’s normal. Each correction improves the system’s accuracy for future documents. Companies typically see significant accuracy gains within the first two to three weeks of regular use, and continued improvement after that as the system encounters more document variations and edge cases.
Lido extracts material lists from engineering drawings, consolidates quantities across multiple sheets, and produces purchase-order-ready output in minutes instead of hours. Start with a free trial—50 pages, no credit card required.
Yes. AI-powered document extraction can identify and extract bill of materials sections from engineering drawings, including piping isometrics, structural steel shop drawings, and other construction documents. Unlike traditional OCR that simply converts images to text, AI extraction understands the structure of these documents and can locate material lists regardless of where they appear on the sheet or how they are formatted. The system captures item descriptions, quantities, units of measure, and material specifications from each drawing.
Construction material lists contain two fundamentally different quantity types: discrete counts (such as 2 elbows or 4 flanges) and linear measurements (such as 40 feet of pipe or 12 feet 6 inches of tubing). The extraction system recognizes both types and preserves the unit of measure alongside each quantity. When consolidating across multiple drawings, counts are summed as whole numbers and lengths are summed as measurements, with automatic conversion between feet-and-inches formats. This prevents errors like adding a count to a length or misinterpreting a footage value as a quantity of items.
Yes. Cross-document consolidation is one of the core capabilities for construction document automation. When processing a package of six drawings, for example, the system extracts material lists from each drawing individually, then matches identical items by their description and specification. All quantities for matching items are summed into a single line item. A 3-inch fitting that appears on four different drawings with quantities of 2, 3, 1, and 4 becomes one line item with a total quantity of 10. This consolidation happens automatically using computed columns, eliminating the need for manual spreadsheet reconciliation.
Engineering drawings for construction typically contain separate sections for shop materials (used in fabrication) and field or erection materials (used during on-site installation). The extraction system uses customizable prompts that instruct the AI to identify and extract only the relevant section. For piping contractors who need shop materials only, the system learns to recognize the shop materials section based on headers, formatting patterns, and document structure, and ignores the field materials entirely. These prompts can be refined over time as the system encounters drawings from different engineering firms with different labeling conventions.
%20(1).svg)
