# TakeoffQS — full content > TakeoffQS is AI-powered takeoff and estimating software for New Zealand and Australian residential material suppliers, quantity surveyors, merchants and builders. It reads residential plan sets and produces measured, quantified takeoffs in minutes, replacing manual measurement in spreadsheets and PDF markup tools, and outsourced QS services that take days per plan. Estimators upload a plan set, review the AI-generated quantities, and push them straight into their merchant catalogue or construction management system. It runs in the browser with no install, and estimators produce their first real takeoff in the first session rather than after weeks of setup. What separates it from general takeoff software is that the computer vision engine is trained on New Zealand and Australian residential plans and the way NZ and Australian estimation desks actually work, not adapted from overseas tools. The model gets stronger with every plan set run through it. TakeoffQS is built in Christchurch, New Zealand as the first product on the BuildFoundry platform. Pre-nail, balance of house, painting and roofing are live today. Product documentation is published at https://help.takeoffqs.com, open to anyone without a login. This file contains the complete text of every published article and the full product FAQ. The link index is at https://takeoffqs.com/llms.txt. --- # Frequently asked questions Source: https://takeoffqs.com/#faq ## Getting started ### What files can I upload? Clear digital NZ residential building PDF plans. ### How do I set the scale? TakeoffQS calibrates every page automatically from the known measurements on your drawings, then puts each one in front of you to verify and approve. If you'd rather set it yourself, you can calibrate any page against a dimension you choose. ### Can I try it without my own plans? Yes — book a demo and we'll walk you through it. If you'd like to see it on your own work, send us one of your plans and we'll run the demo on that. ### Where do I learn how to use it? Our help centre is at https://help.takeoffqs.com — it covers getting started, templates, price books, account setup and exports, step by step. It is open to everyone, no login needed, and we keep it current as the product changes. Customers also get live training from our team during implementation. ## General ### Who is TakeoffQS for? Material suppliers and merchant estimating desks, frame and truss plants, group home builders, residential builders, independent trades and quantity surveyors across New Zealand and Australia. If someone on your team measures plans to price a job, TakeoffQS is built for that work. ### Do I need to install anything? No. TakeoffQS runs directly in your web browser (Chrome, Safari, Microsoft Edge etc.) on a laptop or desktop, so there is nothing to install. ### What does it measure today? The system currently covers pre-nail, balance of house, painting, and roofing (all live). Foundation, concept drawings, aluminium joinery and internal doors, and electrical are coming Q3 2026. Plumbing is coming Q4 2026. ### How fast is it? Most takeoffs are ready in a few minutes. You review and verify the results, tweak if necessary, finalise the bill of materials and export. ### What can I export? Priced material schedules and measurement data export to Excel, CSV, or PDF. You can also export a separate clarifications sheet to send alongside your quote. ### Can I use my own price book and templates? Yes. Import your price book from CSV, Excel, or PDF — TakeoffQS reads your supplier pricing and automatically detects columns like SKU and item code, so measurements land on your own item numbers with no re-keying. You can also set up your own Excel export template so quotes come out in your format. ### How accurate is TakeoffQS? TakeoffQS uses AI computer vision models trained on NZ building plans. A human verification step ensures the output is checked before confirmation. In typical use, the system achieves a very small error margin — however, you should always review the measurements and adjust any lines or counts if required. ### Do I still need to check the measurements? Yes. TakeoffQS is an intelligent assistant — it detects and counts automatically, but you should verify the scale and ensure the lines and counts are correct before exporting. ### What file formats are supported? TakeoffQS accepts clear digital PDF drawings up to 100MB in size. It processes multi-page PDF sets including multi-dwelling plans, but very large files may take longer. ### Will TakeoffQS replace my QS? No. TakeoffQS reduces manual measuring and drafting, but it does not replace the expertise of a professional estimator or quantity surveyor. ## Pricing ### How does pricing work? Pricing is a monthly subscription per organisation, with an allowance of plans included each month. Any plans you run over that allowance are charged per plan. ### How do I get access? Get in touch for a demo and a conversation about what you need. We can also set your organisation up with a pilot, so you can trial TakeoffQS on your own work first. We'll then put together a proposal built around your business, including the right monthly plan allowance. Fill out the contact form or email us at support@takeoffqs.com. ### How does it work for high-volume users? We offer tailored solutions for builders, merchants, and suppliers who process a high number of plans. Contact support@takeoffqs.com to discuss, or call Keegan Payne, our senior sales consultant, on +64 27 463 0680. ## Data & support ### Is my data secure? Yes. Your files are encrypted in transit and at rest. We do not sell your personal contact information. If you require strict data-residency in Australia or New Zealand, contact support@takeoffqs.com. ### How do I get help? Start with the help centre at https://help.takeoffqs.com, which covers every part of the product and is open to everyone without a login. If it does not answer your question, email support@takeoffqs.com during New Zealand business hours. The team aims to reply on the same day. --- # Articles ## Learn more about TakeoffQS Source: https://takeoffqs.com/blog/learn-more-about-takeoffqs/ Author: Nick Latty Published: 2026-08-17 Updated: 2026-09-03 Category: product Tags: getting-started, workflow, quantity-surveyors, builders Where TakeoffQS is at in 2026: live customers, nationwide pilots, a plan-to-quote workflow, H1 compliance, and the team behind it. TakeoffQS is a browser-based construction quantity surveying platform built in Christchurch, New Zealand. It started as a partnership between Base Property Group and Agentic Intelligence, and launched as the first product on the BuildFoundry platform. It's since grown into a standalone company within the BuildFoundry group, with its own team and its own customers. TakeoffQS uses computer vision and AI to speed up the measuring and pricing of a residential construction takeoff. The process stays familiar. The grind goes away. ## Where we're at The product is live and in daily use. Graham Hill Roofing has signed on as a customer, and a number of large New Zealand construction operators are in pilots, finalising agreements, or getting ready to go live. That group spans building merchants, frame and truss manufacturers, building-product manufacturers, group home builders and roofing businesses. We're working with downstream suppliers across the country, from Canterbury to the Bay of Plenty. And the product gets proven on real builds every week through Freedom Built, which puts up 90-odd homes a year in Canterbury. ## Who it is for TakeoffQS is designed for people who plan, estimate, price and supply construction work: - Suppliers turning plan sets into priced material schedules on their own SKUs - Quantity surveyors wanting to spend less time measuring and get more done - Builders preparing accurate quantities before pricing - Trades such as framing, roofing, and painting teams needing reliable counts fast Whether you're pricing one residential build a week or processing 1000 plans a month, TakeoffQS fits into your workflow. ## How it works TakeoffQS follows a four-step flow: 1. **Upload your plans.** Add your PDF plan set. Relevant pages are detected automatically, or you pick the pages you need. 2. **Measure and verify.** AI detects and measures the key building elements. You review, adjust, and verify every number. 3. **Extract your materials.** Formulas turn verified measurements into real material quantities, built to your specs. Wall metres become studs, plates and dwangs at your spacings. Roof areas become tiles, underlay, ridge and flashings. 4. **Price, quote and send.** Quantities land on your own price book items and rates. From there you can build the quote, send packages to suppliers, and share the finished quote with your customer — or export a CSV or priced materials schedule and work outside the platform. We've measured TakeoffQS at 4x the speed of a manual takeoff, with most takeoffs ready in minutes. AI does the heavy lifting, but the professional stays in control. ## What it covers today Pre-nail, balance of house, painting, and roofing are all live now, with more trades landing each quarter as we build new partnerships with experts and train new computer vision models. The plan set itself keeps getting wider. Alongside the architectural pages, the models now read drainage, mechanical, fire, truss, structural, landscaping, site and interior elevation sheets, plus window and door layouts — so more of what you upload is measurable rather than skipped. Duplexes and multi-unit sites model per dwelling, which makes cost per unit a real number instead of a division you do afterwards. And the pricebook now matches a plan's materials to an actual supplier product rather than a rough category. ## Quoting, end to end The takeoff is the hard part, but it isn't the whole job. A governed template library keeps every quote consistent, the bill of quantities prices itself off the plan, and the quote sheet stays editable with material, labour, markup and sell in the footer as you work. Send packages out to suppliers and compare what comes back side by side, then share the finished quote to a client portal where your customer accepts or declines. Plan to signed quote, without leaving TakeoffQS. ## H1 energy compliance Compliance runs inside the same job as the takeoff. The climate zone resolves from the territorial authority, the thermal envelope comes off the plan you've already measured, and R-values are backed by a catalogue with a confirmation step before anything counts. Slab geometry is worked out from Appendix E, including mixed perimeters. You finish with a report that states its working and holds up when someone checks it — no leaving TakeoffQS to prove compliance. ## Why it is different Most takeoff tools were built for overseas markets and later adapted for New Zealand. TakeoffQS was built here first, and its computer vision engine is trained on NZ building plans, including inconsistent drafting styles and layouts. And unlike black-box tools, TakeoffQS is draft-first and human-verified: you can see, check, and edit before anything leaves the platform. ## The team behind it TakeoffQS was co-founded by Cole Askew and Brad Fraser. Cole brings close to two decades in construction, including running multiple building businesses. He's a qualified carpenter and Licensed Building Practitioner (LBP), and he keeps the product grounded in real site and estimating workflows. Brad brings operational discipline and AI execution. A former New Zealand Army officer, he combines strategic delivery with deep AI capability to help teams move faster with confidence. The company has since grown well beyond its founders. Jason Simpson leads the technology as Chief Technical Officer, and Nick Latty runs the business day-to-day as General Manager. Behind them is a 13-strong cross-functional team working out of our HQ in Christchurch: sales, growth, design, marketing, AI engineers, and the annotators who train the computer vision models. The team ships quickly and stays close to builders, QSs, trades, and suppliers, so each release solves a real problem. [Meet the full team here.](/#team) --- ## The AI models behind TakeoffQS Source: https://takeoffqs.com/blog/about-our-ai/ Author: Jason Simpson Published: 2026-07-31 Category: ai-and-technology Tags: ai-models, computer-vision, accuracy, deep-dive The computer vision pipeline, models, and validation that power TakeoffQS. To interpret and understand uploaded build plans, TakeoffQS uses a combination of traditional computer vision techniques, purpose-trained deep learning models, and vision language models. These are designed to handle the variability in real-world plans, where symbols, annotations, and layouts differ significantly between projects and between drafting offices. Rather than relying on a fixed rule set, the system learns patterns directly from training data, then applies geometric and domain checks before producing anything a user sees. This is an update to a post we first published in January. The pipeline has changed enough since then that it was worth rewriting rather than patching. ## System overview At a high level, TakeoffQS processes construction drawings in five stages:
Page classification → Detection & segmentation → Reading the drawing → Geometry & rules → Verification
Each stage addresses a different category of problem. Learned models are used where visual interpretation and ambiguity dominate — deciding what a page is, finding a wall, reading a handwritten-looking dimension string. Classical geometry and explicit rules are used where the answer is determined rather than inferred — turning a wall polygon into a length, a roof outline into an area, a set of segments into a perimeter. The division matters. A model that is 95% right about where a wall is, followed by arithmetic that is exactly right about how long it is, produces a traceable number. A model asked to output the number directly does not. ## Page classification A typical plan set mixes architectural, structural, mechanical and reference drawings, plus title pages, schedules and specifications. Almost nothing downstream works until you know what each page is. Classification runs first, and today it is handled by a vision language model reading the rendered page. It works on visual content rather than file metadata, sheet naming conventions or drafting-office title blocks — those vary too much between practices to rely on, and are frequently wrong or absent. The output routes each page to the appropriate downstream processing. A foundation plan and an elevation get entirely different treatment, so getting this stage right removes a large amount of wasted computation and a larger amount of downstream failure. We're currently bringing a second, much smaller classifier alongside it — a linear model trained on the text extracted from each page, running locally inside our own service. It's not more capable than the vision model; it's faster and cheaper on the pages where the answer is obvious, and it abstains rather than guesses. Where it abstains, or where its confidence sits below a per-page-type threshold we set from measured precision, the page falls through to the vision model as before. It runs in shadow first: a page type only starts answering on its own once its agreement with the vision model has been measured on real traffic and clears the bar we set for that type. That pattern — a cheap model that is allowed to say "I don't know", backed by an expensive one that always answers — is something we now use in several places. ## Detection and segmentation Once a page is classified, TakeoffQS dispatches a specific set of detectors for that page type. There is no single general-purpose model. We run **thirteen detector families**, each trained on its own labelled corpus for one job: external walls, internal walls, roof, foundations, foundation beams, elevations, wall heights, bracing, floor openings, mid-floor, soffits, subfloor piles, subfloor structure. Page-type routing means a floor plan gets the wall and opening detectors, a foundation plan gets pods, beams and piles, and neither pays for the other's inference. The families currently run on two architectures. Five are **RF-DETR-Seg**, a transformer-based segmentation model; eight are **YOLO** variants. That split is historical rather than deliberate, and we're consolidating it — more on that below. For example, on a foundation plan, the segmentation layer detects structural elements including pods and beams. These are often tightly packed and visually similar, with differences in line weight, scale and annotation depending on drafting standards. The model segments individual pods as repeated structural units and identifies beams as elongated connecting elements. Geometric algorithms are then applied to derive ribs and to classify each beam type. Detector output is an intermediate representation, not a result. Nothing a detector produces reaches a user without passing through the geometry and rules layer. ## Reading the drawing Finding shapes is only half the problem. A plan carries a large amount of information in text and symbols that a segmentation model is the wrong tool for: dimension strings, bracing codes and their schedules, roof pitch callouts, fall arrows, downpipe symbols, room names, and the drawing scale itself. We use vision language models for these, but narrowly — cropped to the relevant region of the page at high resolution, prompted for one specific reading, and returned as structured data that is validated against the schema before it is accepted. A pitch that doesn't parse as a pitch is discarded, not guessed at. This is deliberately not "hand the whole plan to a language model and ask for a takeoff". The failure mode of that approach is confident, plausible, wrong numbers with no traceable origin. Scoping each call to one small question against one small crop keeps the output checkable. ## Geometry and rules This is where detections become quantities, and it's the least glamorous and most load-bearing part of the system. The rules layer computes floor area from wall geometry, external wall outline lengths, cladding runs including regeneration of segments hidden behind other elements, lintel sizing and garage type, pitch factors and true roof areas, and room-level wall and paint areas. These are explicit computations over the geometry the detectors produced, at a known scale, with domain constraints applied. They also cross-check. A quantity that can be derived two ways gets derived two ways, and a disagreement is surfaced rather than averaged. ## Where it runs, and who trains it Through the first half of this year we've moved our model training and hosting onto our own infrastructure, and off third-party hosted training and inference. There were two reasons. Control over training — being able to retrain a family when we get better data, rather than when a vendor's platform allows it. And durability — our full training corpus, every dataset version and every trained checkpoint, is now archived in our own storage, so no model we depend on can become unretrainable because a third party changed their retention policy. That archive is not hypothetical insurance; when we pulled it, we found a vendor had already deleted most of the version history for one of our projects. All thirteen families have been ported and parity-tested against their existing behaviour on our own infrastructure. Production is mid-cutover. Alongside that, we've built our own annotation platform for producing training data — the labelled plan corpus that everything above depends on. Owning the annotation step matters more than it sounds: it's the part of the pipeline that determines the ceiling on every model downstream, and it is the part most tied up in a vendor if you don't own it. ## What we're changing next Two pieces of work are in progress. Both are honest works-in-progress rather than shipped features, and we'd rather describe them that way. **Consolidating onto one architecture.** The eight YOLO families are being retrained onto RF-DETR-Seg, the architecture the other five already use. The evidence that this is safe comes from a direct comparison: two of our families are trained on the same dataset, one on each architecture, and they score within a point of each other. Architecture is not what's limiting us — sample count is. Consolidating gets us to a single serving stack and a single training path, which makes every subsequent improvement cheaper to apply across the fleet. **Feeding the model the geometry it's currently throwing away.** Today our detectors read plans as photographs. We rasterise a vector PDF to an image and ask a model to recover line work from pixels, after discarding the exact line geometry the PDF gave us for free. At the resolution we work at, a wall is one to two pixels wide; as a vector primitive, it's exact. The obvious shortcut — read the geometry directly and skip the model — doesn't survive contact with real plans, and we've tested that against our own corpus rather than assuming it. Making that geometry available to the model instead is the work we're in the middle of. ## What this doesn't change None of the above changes the fundamental position we've written about [elsewhere](/blog/the-real-metrics-for-ai-takeoffs): TakeoffQS produces a draft takeoff for supported residential scopes, which a qualified professional verifies, corrects and takes responsibility for. Better models make the first draft closer and the verification faster. They don't remove the verification step, and we're not building toward a version where they do. --- ## Local plans, local accuracy: why it matters for merchants Source: https://takeoffqs.com/blog/local-plans-local-accuracy-merchants/ Author: Duane Smithson Published: 2026-07-13 Updated: 2026-08-06 Category: industry Tags: nz-specific, merchants, white-label Building merchant estimating NZ depends on local accuracy. Here is why NZ-trained takeoffs put estimates in front of your customers in minutes, with a human sign-off on every number. **In short:** Building merchant estimating NZ depends on the model having learned on New Zealand plans. Local conventions mean local accuracy, and that is what protects the estimate you hand your customer. If you run estimating for a building merchant, you already know the stakes. Every estimate you send is a promise. Building merchant estimating NZ lives or dies on whether the numbers match what gets built, and a model that has not learned on NZ drawings has nothing to go on when Kiwi conventions differ. Local plans need local accuracy. ## Why NZ plans are their own thing A New Zealand residential plan is not a generic drawing. We have our own framing conventions, our own bracing notation, our own way of calling up cladding, and timber sizing that follows NZS standards. A model that has only ever seen American or European plans has not learned any of that. It guesses, and a guess on a takeoff becomes a wrong line on a customer estimate. TakeoffQS is trained on New Zealand residential plans. Our computer vision model has read the way NZ architects and draughtspeople actually annotate, so it picks up detail that only makes sense in a NZ drafting context. That is the difference between merchant takeoffs you can stand behind and ones you have to re-check by hand. ## What local accuracy means for the merchant Merchants sit in a tricky spot. You are not the builder, but you are the one whose estimate the builder uses to make a call. If your supplier estimating is off, you either lose the job on a number that was too high, or you wear the gap on a number that was too low. Neither is a good day. Local accuracy gives you: - Material lists that match NZ product ranges and pack sizes - Cladding and lining takeoffs that read local call-ups correctly - Framing quantities aligned to the timber sizing your customers actually order - Fewer manual corrections before the estimate goes out the door When the takeoff starts right, the estimate starts right. ## White-label estimates that carry your name Plenty of merchants want to put estimates in front of customers under their own brand. That is exactly the point of white-label estimates: the speed and accuracy of AI-powered takeoffs, wearing your logo, not ours. But a white-label estimate only helps you if the number underneath is sound. Slap your brand on a sloppy takeoff and you own the mistake. This is why NZ plans accuracy matters more for merchants than almost anyone. You are lending your reputation to every estimate. The model has to be right on local drawings before you would ever hand the output to a builder. ## How TakeoffQS keeps it local Here is the practical side. You upload a plan, TakeoffQS reads it with our computer vision model, and a draft takeoff comes back in minutes instead of hours. Then a human on your team approves it. Speed from the machine, sign-off from a person who knows your customers and your stock. The workflow looks like this: - Upload the customer's plan set. - TakeoffQS produces a draft takeoff and estimate. - Your estimator reviews and adjusts to your pricing. - The white-label estimate goes to the customer under your brand. Every step keeps a human in control of the final number. The model does the grunt work of measuring, which is the slow part, and your people do the judgement, which is the part that needs a human. ## Why local training matters A demo is not a plan set. Build costs here have been climbing, margins are tight, and customers notice when an estimate is off. What you need to know before you commit is whether the model has learned on the drawings your customers actually send you — because a Christchurch plan is not a Texas one, and a model can only read what it was trained to read. Local training is not a marketing line for us. It is the reason the output is usable straight away. We are based in Christchurch, part of the BuildFoundry platform, and TakeoffQS is trained on New Zealand residential plans and gets better with every plan our customers run through it. If you want the detail on how that accuracy is measured, our note on [the real metrics for AI takeoffs](/blog/the-real-metrics-for-ai-takeoffs) walks through it. ## Key takeaways - Building merchant estimating NZ needs a model trained on NZ plans. - Local accuracy means material lists, framing and cladding match what your customers actually order. - White-label estimates only help if the takeoff underneath is sound. - Speed comes from the model, the final number stays under human control. - Tight margins and rising costs make accuracy worth more, not less. ## FAQ **Why can't I just use any AI takeoff tool?** Not all AI is the same, and the difference matters here. Most people picture a chatbot, a large language model, and that is the wrong tool for a takeoff. An LLM will give you an answer whether or not it is right, and it will sound completely sure of itself while doing it. On a set of building plans, that confident guessing is exactly what you do not want near a number with your brand on it. TakeoffQS is an entirely different beast. It is a purpose-built computer vision model, trained on NZ plans, that measures what is actually on the page rather than predicting what sounds likely. That is why it reads framing notation, bracing, cladding call-ups and timber sizing the way we build here, instead of producing takeoffs you have to fix by hand. **What does white-label actually mean for my merchant business?** It means the estimate goes to your customer under your brand, not ours. You get the speed of AI-powered takeoffs with your name on the output, which keeps the customer relationship yours. **How long does an estimate take?** TakeoffQS reads a plan and returns a draft takeoff in minutes rather than hours. Your estimator then reviews and adjusts it before it goes out, so the time saved is on measuring, not judgement. **Does a human still check the numbers?** Yes, always. The model measures, your team approves. You get the speed without blindly trusting a machine on a number that has your brand attached to it. Want to see how local accuracy looks on your own plans? Send us a plan set, or book a demo, and we will show you what merchant takeoffs look like when the model already knows New Zealand drawings. --- ## The QS shortage is real, and it is not going away Source: https://takeoffqs.com/blog/qs-shortage-nz/ Author: Cole Askew Published: 2026-06-22 Updated: 2026-08-06 Category: industry Tags: nz-specific, estimators, opinion The quantity surveyor shortage NZ faces is not easing. Here is how to do more with the estimating expertise you already have on the team. **In short:** The quantity surveyor shortage NZ faces is structural, not temporary. You cannot reliably hire your way out, so the fix is doing more with the estimating expertise you already have. If you have tried to hire an experienced estimator lately, you already know. The quantity surveyor shortage NZ is dealing with is not a blip. Quantity surveying has been on Immigration New Zealand's long-term skill shortage list, which is the government's own way of saying we do not have enough of these people and cannot train them fast enough. This is a structural problem, and pretending it will sort itself out is a plan for falling behind. ## Why the shortage is structural A good quantity surveyor takes years to train. You cannot conjure one when a job lands. The pipeline of new estimators coming through has not kept pace with the work, experienced people are retiring, and the ones still working are stretched thin. That is what a structural shortage looks like: demand outruns supply, year after year. This is not unique to estimating. The wider construction labour shortage has been biting across trades. But estimating is a special kind of bottleneck, because nothing prices and nothing proceeds until the takeoff is done. When you cannot find estimators, the whole job queue backs up. ## What the shortage actually costs you The QS shortage New Zealand businesses feel does not show up as a single line item. It shows up as: - Jobs you turned down because you had no capacity to estimate them - Estimates that went out late and lost to a faster competitor - Your best estimator buried in measuring instead of doing high-value work - Burnout and turnover, which makes the shortage worse for everyone Every one of those is margin walking out the door. And with build costs climbing, the cost of a slow or missed estimate is higher than it used to be. ## You cannot hire your way out The instinct is to hire more estimators. Good luck. That shortage exists precisely because those people are not available, and importing them is slow and uncertain. Even when you find someone, you are competing with everyone else who is also short. Throwing job ads at a structural shortage does not fix it. So the question changes. Instead of "how do I find more estimators", it becomes "how do I get more out of the estimators I already have". That is a problem you can actually solve. ## Doing more with the expertise you have This is where the maths shifts in your favour. Most of an estimator's day is not judgement. It is measuring. Counting, scaling, totalling, the slow repetitive work of getting numbers off a plan. That part does not need years of QS training. It needs accuracy and patience, which is exactly what a machine is good at. Here is the practical approach: 1. Let TakeoffQS read the plan and produce a draft takeoff in minutes. 2. Your estimator reviews and approves it, applying judgement where it counts. 3. The expertise you are short of gets spent on the high-value part, not the grunt work. 4. The same estimator now covers more jobs without working longer hours. That is how you grow estimating capacity without hiring. You stop spending scarce expertise on tasks that do not need it, and your senior people get their time back for the work that actually needs a trained eye. ## Why TakeoffQS fits the shortage TakeoffQS reads building plans with our computer vision model and produces draft takeoffs and estimates in minutes, which a human then approves. It is built for exactly this problem. The machine does the measuring, the slow part, and your experienced people do the judgement, the part that is genuinely scarce. It does not replace your QS. It gives them their time back. The measuring comes off their desk, so the same experienced estimator covers more jobs without working longer hours. In a market where you simply cannot hire the next estimator, that is the difference between taking the work and turning it away. We are based in Christchurch, part of the BuildFoundry platform, and TakeoffQS is trained on New Zealand residential plans and gets better with every plan our customers run through it. For why that local training matters, see [why AI takeoff software for NZ has to be trained on NZ plans](/blog/ai-takeoff-software-nz). ## Key takeaways - The quantity surveyor shortage NZ faces is structural and not easing. - You cannot reliably hire your way out of a shortage this size. - The real cost is missed jobs, late estimates and burnt-out staff. - Most estimating time is measuring, not judgement, and measuring can be done by machine. - AI-powered takeoffs let your existing estimators cover more work. ## FAQ **Is the QS shortage really that bad?** Yes. The gap between demand and supply is structural, not a short-term blip, and it takes years to train a good estimator, so it is not easing quickly. **Will AI replace my quantity surveyor?** No. It takes the slow measuring work off them and keeps the final number under human control. The point is to let your scarce expertise cover more jobs, not to remove the person whose judgement you depend on. **How does this increase my estimating capacity?** By removing the repetitive measuring from your estimator's day. TakeoffQS produces the draft takeoff in minutes, your estimator reviews and approves it, so the same person handles more jobs without longer hours. **We are a small team. Does this still help?** Especially then. Small teams feel the shortage hardest because losing one estimator hurts more. AI-powered takeoffs stretch the people you have, which matters most when you have few of them. Short on estimating capacity and feeling the shortage? Send us a plan set, or book a demo, and we will show you how much more your team can cover. --- ## Prenail in 2026: the quote that goes out first wins Source: https://takeoffqs.com/blog/prenail-2026-quote-first-wins/ Author: Keegan Payne Published: 2026-05-27 Updated: 2026-08-06 Category: industry Tags: prenail, quoting, speed, nz-specific, opinion In NZ prenail, the first credible quote wins the job — and usually the next three. Why compressing tracing time is the edge in 2026. If you're running an NZ frame and truss plant right now, you already know the squeeze. Residential is recovering unevenly. Builders are quoting more carefully and committing later. Pipelines are noisier than they were two years ago. The plants winning work in 2026 are the ones whose quotes land first, accurate, and signed off by someone the builder trusts. That last part is where most plants are leaking jobs they should have won. ## Where the time actually goes Walk into any prenail estimating office and watch a senior QS work a plan set. Hours go to tracing. Counting studs. Reading elevations. Measuring rakes. Typing numbers into a takeoff package or someone's in-house spreadsheet so the pricing engine has something to chew on. That work is data entry, and it creates the all-too-familiar backlog of jobs sitting on the desk. Those 4 to 6 hours of measurement and data entry decide whether the real conversations with your client, the ones where you shape and refine the price to actually win the job, happen that day or a week later. In 2026, a week down the track is too late. ## The job goes to the first credible quote Builders go with whoever quotes first with confidence and stands behind the number. Plant loyalty barely registers anymore. Trust and credibility have to stay front of mind for leadership, and that's hard to build when you can't even guarantee when you'll get to a job. We've heard about this playing out on real jobs. Plan set goes out to three plants Monday morning. One plant turns a quote around Tuesday afternoon. The other two are still tracing on Thursday. By the time their quotes land, the builder is already in conversation with the first plant about variations, lead times, and where the price needs to sit to fit the client's budget and get the job over the line. Relationships in this industry are built on trust. If your client can trust you to deliver a quote quickly, your credibility builds and price has very little to do with it. That's the part most plants underestimate. The quote that goes out first wins the job, and it usually wins the next three jobs from that builder too, because the builder now knows exactly where to send the next plan set. ## So what changes in 2026 Two things, and they have to happen together. First, the tracing work has to compress. A multi-page plan set should become a draft takeoff in minutes. Framing, foundations, roofing metrics, cladding surface areas, the parts that are pure measurement get done by a computer that doesn't get tired at 4pm on a Thursday. Second, the estimator has to stay in charge of the workflow. Every measurement gets a verify-and-confirm step. The estimator's name is on the export. The audit trail sits behind every number. When the builder asks "where did that come from," the QS can show them, line by line. That's the shift worth getting right. The estimator stops being the operator typing numbers into a spreadsheet and starts being the auditor signing off on a draft the machine produced. Same person. Same seat. Better use of the experience they were hired for. ## What it does for a prenail plant The math gets interesting fast. If your senior QS spends 2 to 3 hours per plan taking measurements, and that work compresses to 30 minutes of verification, you've got 1.5 to 2 hours back per plan. That's another quote out the door the same day. Across a year, it's the difference between turning work away and bidding for more of it. The prenail plants we're working with use this to stop losing jobs to slow turnaround and to take on the volume they already have to turn down. The senior QS keeps doing what they were hired for. Judgment. Builder relationships. The build-it-before-it's-built thinking that a computer can't fake. That's the part we care about most. A plant where the estimators feel elevated by the tool adopts it. A plant where they feel threatened by it doesn't, regardless of what the GM signs. We've designed the workflow around that, on purpose. ## A quick note on what this isn't It isn't engineering. It isn't compliance sign-off. It isn't a bracing or structural adequacy calculation. It isn't a black box that pushes a number out the other end with no one to ask. It's a workflow tool. The trace work compresses. The judgment work stays exactly where it belongs. ## Where this is going The next 18 months in NZ prenail will separate the plants that quote in hours from the plants that quote in days. We're building TakeoffQS on real NZ residential plan sets, working alongside prenail plants that look a lot like yours. Most plants will get there eventually. The real question is whether you'll be the first plant in your region to do it, or the third. If you want to see what plan-to-quote in minutes looks like on real NZ residential plan sets, including the verification step the estimator owns end to end, book us in for a demo. --- ## Why AI takeoff software for NZ has to be trained on NZ plans Source: https://takeoffqs.com/blog/ai-takeoff-software-nz/ Author: Nick Latty Published: 2026-04-22 Updated: 2026-08-06 Category: industry Tags: nz-specific, ai-models, accuracy, opinion Training data lineage decides whether AI takeoff software reads NZ drawings properly. Here's what breaks when it doesn't (scale, symbols, terminology, layout) and why it matters for NZ residential estimating. AI takeoff software is only as good as the plans it's learned from. Most of the AI estimating tools sold into NZ were trained somewhere else, on drawings that don't look like ours. That matters more than people think. When we started building TakeoffQS, we set one rule early. We'd train our models on NZ residential plans, keep training them on NZ residential plans, and never shortcut that with a generic overseas dataset. Every framing detail, every scale bar, every symbol the model sees comes from a real plan set produced by an NZ architect or drafter. Six months in, that decision already defines what the product does well and where overseas takeoff software falls over. Here's what breaks when the training data doesn't match the market. ## Scale defaults built for the wrong continent A lot of overseas-built takeoff software was trained on plans from other markets. That means it learned on different sheet sizes, different scale conventions, and different layout standards than NZ drafters use. Drop an NZ A3 1:100 plan into one of those tools and the scale detection is working across a format it wasn't trained on. The estimator ends up doing calibration work the tool should've done for them. Our scale detection was trained on A3 and A1 NZ sheets using NZ scale conventions (1:50, 1:100, 1:200). It looks for scale bars and title blocks the way they sit on NZ plans, not the way they sit on American or British ones. Small difference on paper, big difference in whether the first draft lands trustworthy or needs manual rework. ## Symbol libraries from another country Takeoff symbols aren't universal. NZ drafters draw strip bracing, garage doors, fall arrows, and elevation openings differently from their US or UK counterparts. A model trained on overseas plans will misclassify the symbols or skip them entirely. TakeoffQS was trained on NZ residential plans end-to-end. The models pull the things NZ suppliers actually need to count and measure: - Floor plans: exterior walls, interior walls, wall widths, windows, doors, garage doors, floor openings, strip bracing - Roof plans: outline, catchments, valleys, gable ends, soffits, and fall arrows for pitch and direction - Foundation plans: perimeter, pods, ribs, beams - Elevations: cladding surfaces, doors, windows, facade regions When a plan lands in TakeoffQS, the model already knows what it's looking at. ## Trade terminology built for a different market Ask most overseas AI estimating tools for a "balance of house" takeoff and they won't know what you mean. BoH is a NZ and AU supplier term. It's the structural and surface quantities a material merchant quotes outside the prenail package: foundations, roofing metrics, cladding, soffits, painting, internal linings. Overseas-built tools don't carve the work up that way, because their customers don't use the term. Ours does, because our customers asked for it. Same story with prenail. Overseas takeoff software handles it as generic panelised framing. Our prenail output is shaped around what NZ frame and truss yards actually price: exterior and interior wall lengths, window and door openings, and the counts that flow straight into a yard's pricing spreadsheet. ## Plan layouts trained on the wrong drawings NZ architects lay out plan sets differently from US or UK ones. Title blocks sit in specific positions. Page ordering follows a loose convention: site, floor, elevations, sections, details. Revision clouds and markup follow NZ drafting conventions. Overseas AI takeoff software struggles with this silently. It processes the pages, but the part that sorts and routes them was trained on a different tradition, so it'll misroute an elevation as a section or skip a critical detail page entirely. The estimator doesn't always notice until the numbers come out wrong. We trained our system on how NZ architects lay out plan sets. It knows where the site plan sits, where the floor plan sits, where the elevations sit, and it pairs those pages up so detections on each page are actually comparing like with like. ## Why the training data matters Models can be cloned. The plans they learn from have to be earned. TakeoffQS is trained on New Zealand residential plans, and gets better with every plan our customers run through it. Christchurch townhouses, Queenstown holiday homes, Hamilton subdivisions — the corpus is built from the drawings NZ estimators actually work on. Six months of this. New Zealand building plans, and nothing else. Growing every week. If you're evaluating AI takeoff software for NZ residential work, ask one question. What was the AI actually trained on? If the answer is "global construction drawings" or "tens of thousands of plans" with no NZ specificity, you already know what'll break. ## Same method, next market We're already training on Australian and US plans. The UK and Canada follow from there. Each market gets its own set of training plans, pulled from real drawings produced by local architects and drafters. We're not porting the NZ model across and hoping it holds. When we land in Australia, the models will know Australian drafting conventions and trade terminology. When we land in the US, they'll know ANSI sheet sizes, imperial defaults, and US framing conventions. That takes longer than training once and selling everywhere. It's how we keep producing AI takeoff software that reads plans the way local estimators actually expect. Elite products are made this way. Patiently, on the right data, for the people who have to live with the output. --- ## The real metrics for AI takeoffs Source: https://takeoffqs.com/blog/the-real-metrics-for-ai-takeoffs/ Author: Nick Latty Published: 2026-02-01 Category: ai-and-technology Tags: accuracy, ai-models, workflow, opinion Why a single accuracy percentage is the wrong question — and what to measure instead. When people hear "AI takeoffs," the first question is often: "How accurate is it?" It's a reasonable question, but there's more to unpack here. In residential construction estimating, you need speed to respond while jobs are still winnable, control so you can stand behind your quantities, and accountability so responsibility stays with the qualified professional. The better question is: "Can this system produce a fast first draft that I can verify, price and confidently export?" There's a fundamental difference between automation that replaces judgement and assistance that removes tracing work. ## Accuracy is a workflow Takeoffs fail when the scale is wrong, when someone measures an outdated plan revision, when a symbol gets misinterpreted, or when a detail is missed or double-counted. These are workflow problems, not model problems. Chasing a single accuracy percentage misses the real opportunity: reduce time-to-first-draft dramatically, then make verification fast and explicit. That's how you move quicker and stay safe. ## Verified exportable quantities & prices, fast Estimators and QS professionals get paid for judgement — deciding what matters, catching what's missing, understanding scope boundaries, and producing quantities the business can act on. The bottleneck is the manual drafting and tracing time that delays that judgement. A good system produces a first-pass draft quickly so you're not starting from zero, makes verification mandatory and straightforward so you keep control, and exports cleanly into existing workflows so it actually creates value. ## The real metrics to track If you want to evaluate AI takeoffs the way an estimating team experiences them, measure outcomes that map to throughput and accountability: - **Time to first draft** — how long from upload to a usable first-pass takeoff. - **Time to verified, exportable quantities** — how long from first draft to "approved" and ready to export. - **Verification effort** — how much human time it takes to review and correct (the goal is to make this fast and explicit). - **Edit rate (a healthy signal)** — if users are verifying and editing, the workflow is working. Zero edits can mean either perfection or blind trust. - **Export rate** — drafts that never get exported aren't value, just previews. ## What "human-in-the-loop" actually means "Human-in-the-loop" often gets used as vague reassurance. For takeoffs, it should mean something specific: - The output is explicitly a draft. Not a certification, compliance statement, or something you blindly accept. - The user must verify. The qualified professional confirms scale, reviews detections, and edits anything needed. - The user stays responsible. The workflow keeps accountability with the estimator or QS, not with the software. TakeoffQS is built around this pattern: fast first-pass takeoffs for supported residential scopes, followed by required verification, then export to CSV / Excel. We're explicit about what we provide — draft takeoffs for supported scopes — and what we don't: engineering certification, compliance statements, or bracing calculations. ## Conclusion The point isn't to pretend estimating can be automated end-to-end. It can't — not safely, not consistently, and not in a way a professional can stand behind. What does work is a workflow where AI removes the slowest, most repetitive part (first-pass measurement), and the estimator keeps the part that actually matters: confirming scale, reviewing what's been detected, correcting edge cases, and owning the final export. ---