Automating takeoff: the QS's new toolbox
The Quantity Surveyor produces a project's quantitative takeoff: the detailed list of every wall, every door, every m² of concrete, with a unit price and a total. It is the profession most impacted by AI automation in 2026 because it is the most predictable (clear rules, structured output files) while still carrying the technical responsibility for the cost. This article draws on feedback from 40 QS firms — UK, France, Switzerland, Belgium — to describe the toolkit actually deployed in 2026 and the measured impact on productivity and the role of the QS.
Metrological OCR: reading dimensions on the plan
First essential tool: specialised OCR for automatic reading of dimensions ("3.40 m", "1.20", "0.83 × 2.04"). General-purpose solutions (Tesseract, AWS Textract) fail on plans because of text rotation, non-standard CAD fonts and mixed units. Metrology tools (FloorScan, Bluebeam Revu with its Quantity Link plugin, On-Screen Takeoff) integrate a fine-tuned OCR that hits 97-99% accuracy on standard architectural dimensions.
Typical use: a QS who estimated 80 plans/month by reading dimensions manually moves to 200 with automatic OCR. The real gain is not only reading time (5 seconds vs 30 seconds per dimension) but the absence of transcription errors. Human reading errors (1.40 read for 1.80, commas confused with periods) represented 0.5-1% of entries and each one cost $5-$30 in wrong-cost penalty. OCR brings this rate below 0.1%.
AI object detection: doors, windows, surfaces
Second brick: AI detection of discrete architectural objects. On a 1,200 m² office floor plan, manually counting 87 doors, 145 windows, 38 fire-rated partitions takes 90 minutes; an AI model does it in 30 seconds with 95% accuracy, the QS validates the 5% of errors in 10 minutes. The result comes out directly as a structured Excel BOQ: one row per element, type, dimensions, unit price to fill.
Integration with internal price-item libraries is the next step: 2026 tools (FloorScan via API, Bluebeam Revu via XML) let you automatically map each detected type ("internal single door 0.83 × 2.04") to an item in the in-house catalog with its unit price. The BOQ comes out priced, not just quantified. For a QS, that is the difference between 30 minutes of data entry and 30 seconds of validation per project.
ERP integration: Batigest, Sage, RIB iTWO
Takeoff doesn't live in isolation: it feeds the quote, the quote feeds the contract, the contract feeds scheduling and billing. A QS firm's maturity in 2026 is measured by the depth of this integration. Three observed patterns: (1) native plugin integration (Batigest and Onaya offer a Bluebeam plugin, RIB iTWO has one with Plan Grid); (2) REST API integration (FloorScan, some Revu extensions) that automatically pushes quantities to the ERP via webhook; (3) manual integration through Excel export + VBA macro (most widespread, least robust).
Firms that invested in native API integration report a 60-70% cut in cumulative entry time across the full cycle (from plan received to quote sent). Firms staying on Excel + VBA gain only on the takeoff pass (~30% of total time), not on the downstream — hence more modest ROI.
Audit and traceability: the new responsibility
With automation, the QS's role shifts from calculation to control. The core skill is no longer counting fast but guaranteeing the final accuracy of the takeoff delivered to the client — which may now be partly AI-generated. This imposes a complete audit trail: for every BOQ line, who (human or AI) produced it, when, on what basis, with what confidence. 2026 tools are starting to integrate these logs (FloorScan generates a timestamped session PDF for each detection, Bluebeam Revu has its Studio module for change tracking).
In practice, organised firms pass every AI takeoff through an 8-point checklist before signing: at least 95% of expected elements present, dimensions consistent with known ranges (door ~0.8 m, window ~1.2 m), 100% visual validation of uncertain items (confidence < 80%), calibration recheck if outlier values, cross-check against the previous quote from same client (alert if delta > 10%), e-signature by the responsible QS, archiving of the AI session, transmission of the audit PDF with the BOQ.
The profession's evolution: from calculation to advice
The QS profile recruited in 2026 has changed. Where five years ago firms looked for pure quantitative takeoff expertise (BIM, spreadsheets, Batigest), they now want a triple skill set: (1) trade technical (still able to do a manual takeoff to validate the AI), (2) tool mastery (able to configure a Bluebeam-Batigest integration, read an AI session log), (3) client advisory (able to explain a price gap between two variants, walk the owner through trade-offs). Specialised schools (ESTP in France, CFC in Switzerland) updated their curricula in 2024-2025 to integrate the tool component.
The profession is not disappearing — on the contrary, QS demand grew 12% in 2025 according to RICS UK — but its nature is changing. You gain added value (less data entry, more strategy), you lose repetition (juniors who learned the trade by doing 200 manual takeoffs a year no longer do, which creates a knowledge-transmission challenge).
Automating takeoff is not a horizon, it is the present. QS firms that engaged the transition in 2023-2024 process three to five times more projects with the same headcount in 2026, while dedicating more time to strategic advice. Those who waited find themselves with 2018-baseline productivity in 2026 and margins under pressure. The entry door is always the same: start with a single OCR or AI detection tool, industrialise it for 3 months on 30 real projects, measure the gains, then extend the integrated ecosystem (ERP, e-signature, archiving). The QS profession remains as necessary as before — it now operates at the strategic floor, no longer at the data-entry floor.