Construction estimating & field management
Do quantity takeoff, estimates, RFIs and daily logs in construction software.
20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.
- Typical applications
- Procore · Bluebeam Revu · PlanSwift
- Class GUI-need label
HY· Hybrid
Task proposals · 20
Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.
R05.13.01.06-T01 Calibrate a plan sheet and measure interior partitions
Original instruction
In Bluebeam Revu on Windows, open input/plans_rev3.pdf sheet A-101 (raster scan, no vector data). Calibrate the scale to 1/8" = 1'-0" using the dimensioned grid line, then measure every interior partition (exclude exterior walls and casework) with polylength markups on layer 'Partitions' and save. Write the total length to output/partitions.txt.
- Platforms
win· Windows- Applications
- Bluebeam Revu
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
path· Continuous pathgeom· Precise geometrynav· Pan/zoom/scrollvis· Visual judgment- Work product
WP10· Annotations- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Parse saved PDF annotations: scale calibration on the page; polylength markups on layer; total within 2% of reference 412 ft; markup geometry overlaps reference walls (buffered IoU >= 0.7). Raster sheet blocks vector extraction; Revu scripts disabled.
Proposed partial credit
20% calibration; 50% total length accuracy (full within 2 pct, linear to 8 pct); 20% geometry overlap; 10% layer and file saved.
R05.13.01.06-T02 Count receptacles across two electrical sheets
Original instruction
In PlanSwift (Windows), open project Clinic_TI with sheets E-201 and E-202. Count all duplex receptacles (exclude GFCI, quad and floor boxes, which use similar symbols) with a count item 'Duplex Receptacle' and export the takeoff report to output/receptacles.xlsx.
- Platforms
win· Windows- Applications
- PlanSwift · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
vis· Visual judgmentclick· Point & clicknav· Pan/zoom/scroll- Work product
WP10· Annotations- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Report total vs reference 87 (exact for full credit, linear to +/-5); count point positions match reference symbols within 0.5 in at sheet scale (precision/recall). Symbols are raster, so text search cannot count them. PlanSwift's plugin/scripting SDK is disabled and the VM has no other runtime, so takeoff items must be created in the PlanSwift UI.
Proposed partial credit
60% count accuracy; 30% precision/recall of points; 10% report export.
R05.13.01.06-T03 Measure slab area and price it with waste
Original instruction
In Bluebeam Revu, on sheet S-101 of input/structural.pdf, measure the Level 1 slab-on-grade area (exclude the two depressed slab areas hatched on the plan, which are priced separately). Create a custom column 'Cost' = area x $8.75 x 1.05 and export the markup summary to output/slab.csv.
- Platforms
win· Windows- Applications
- Bluebeam Revu
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
path· Continuous pathgeom· Precise geometryvis· Visual judgment- Work product
WP10· Annotations- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Area markup(s) IoU >= 0.95 with reference polygon minus depressions; total area within 1%; custom column formula present; CSV cost equals area x 9.1875. Revu scripts and batch automation are disabled, so markups must be made in the Revu UI.
Proposed partial credit
45% area within 1 pct; 20% exclusions handled; 20% cost formula; 15% CSV.
R05.13.01.06-T04 Build an estimate summary with assemblies and cost codes
Original instruction
In PlanSwift, project Clinic_TI: take off flooring (VCT, carpet tile), ACT ceilings and base on A-101/A-102 per finish schedule input/finishes.pdf, attach assemblies with unit costs from input/unit_costs.xlsx, and produce an estimate summary by the 4 cost codes in input/codes.txt. Export to output/estimate.xlsx; total must be within 1% of reference.
- Platforms
win· Windows- Applications
- PlanSwift · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
path· Continuous pathvis· Visual judgmentclick· Point & clickmulti· Cross-app- Work product
WP02· Records/table- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Export parsed: quantities per item within 2% of reference; assemblies linked to correct items; cost-code subtotals and grand total within 1%. PlanSwift's plugin/scripting SDK is disabled and the VM has no other runtime, so takeoff items must be created in the PlanSwift UI.
Proposed partial credit
40% quantities; 20% assemblies; 25% code subtotals; 15% total within 1 pct.
R05.13.01.06-T05 Compare drawing revisions and cloud the changes
Original instruction
In Bluebeam Revu, compare input/A-201_rev2.pdf and input/A-201_rev3.pdf (Compare Documents or Overlay). Place a revision cloud with a callout on rev 3 around each real design change (ignore title-block and plot-date differences), number them, and export the markups list to output/changes.csv.
- Platforms
win· Windows- Applications
- Bluebeam Revu
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
vis· Visual judgmentpath· Continuous pathnav· Pan/zoom/scroll- Work product
WP10· Annotations- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Clouds matched to reference change regions (IoU >= 0.4 each); recall and precision over 9 real changes; no clouds on title-block noise; CSV row count equals clouds. Revu scripts and batch automation are disabled, so markups must be made in the Revu UI.
Proposed partial credit
50% recall of changes; 25% precision (no noise clouds); 15% callouts numbered; 10% CSV.
R05.13.01.06-T06 Submit an RFI with a marked-up drawing
Original instruction
In Procore (sandbox project 'Harbor Clinic'), create an RFI from the superintendent's note input/field_issue.txt (duct conflicts with beam at grid C/4): question text, drawing reference A-301, attach input/conflict_markup.pdf, assign to the architect, distribution per input/rfi_contacts.txt, due date 7 calendar days per contract.
- Platforms
web· Any OS (web)- Applications
- Procore
- Interfaces
brow· Browser- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Procore sandbox RFI: subject/question include grid C/4 and beam conflict, drawing link A-301, attachment hash, assignee and ball-in-court architect, distribution list exact, due date = today + 7.
Proposed partial credit
30% question content; 20% drawing link; 20% attachment; 30% assignee, distribution and due date.
R05.13.01.06-T07 Enter a daily log from superintendent notes
Original instruction
In Procore (sandbox), complete today's daily log for 'Harbor Clinic' from input/super_notes.txt: manpower by company and hours, weather observations, 3 deliveries, a 2-hour delay due to inspection with cause, and attach the 4 site photos to the photo section.
- Platforms
web· Any OS (web)- Applications
- Procore
- Interfaces
brow· Browser- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
type· Text & formsclick· Point & click- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Daily log entries via grader: manpower rows (company, workers, hours) exact; weather; deliveries; delay entry with duration and cause; 4 photos attached to the date.
Proposed partial credit
35% manpower; 15% weather; 20% deliveries; 30% delay and photos.
R05.13.01.06-T08 Build a submittal register from the spec index
Original instruction
In Procore (sandbox), create the submittal register for divisions 08 and 09 from input/spec_index.pdf: one submittal per required item (product data, shop drawings, samples) with spec section, type, responsible subcontractor from input/subs.xlsx, and the 2-step approval workflow (GC review then Architect).
- Platforms
web· Any OS (web)- Applications
- Procore · Excel
- Interfaces
brow· Browser- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
type· Text & formsclick· Point & clickmulti· Cross-app- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Submittals set equals key (section x type); responsible contractor; workflow steps and approvers exact on each.
Proposed partial credit
50% submittal items; 25% responsible subs; 25% workflows.
R05.13.01.06-T09 Turn a change event into a priced change order request
Original instruction
In Procore (sandbox), create change event 'Owner adds exam room 112' from input/owner_directive.pdf, add line items from the 3 subcontractor quotes in input/quotes/ (correct cost codes, markups per input/prime_contract.pdf: 10% OH&P on subs, 5% bond/insurance on total), and generate a prime contract change order request.
- Platforms
web· Any OS (web)- Applications
- Procore
- Interfaces
brow· Browser- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
type· Text & formsclick· Point & clickmulti· Cross-app- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Change event line items with cost codes and amounts per quotes; PCO with markups computed per contract to the cent; linked to change event.
Proposed partial credit
35% line items; 20% cost codes; 35% markups and total; 10% PCO linked.
R05.13.01.06-T10 Reconcile reviewer markups in a shared review session
Original instruction
In Bluebeam Revu, join Studio Session 'Harbor-DD-Review' (test Studio). Three reviewers left 46 markups on the DD set. Set status 'Accepted' or 'Rejected' per input/review_decisions.xlsx (matching by markup subject and page), reply to rejected ones with the reason, then export a markup summary CSV to output/review.csv.
- Platforms
win· Windows- Applications
- Bluebeam Revu · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicknav· Pan/zoom/scrolltype· Text & formsmulti· Cross-app- Work product
WP10· Annotations- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Session record export: statuses per markup equal key; replies on rejected markups contain reasons; CSV complete. Status changes must be attributed to the agent's Studio user. Studio API credentials are withheld and Revu scripting is disabled, so status changes must be made in the Revu session.
Proposed partial credit
55% statuses; 25% replies; 20% CSV.
R05.13.01.06-T11 Create sheet bookmarks and hyperlinks for a drawing set
Original instruction
In Bluebeam Revu, open the 64-page input/drawing_set.pdf, create page labels and bookmarks from the sheet numbers and titles in each title block, and auto-hyperlink the sheet and detail callouts. Save as output/drawing_set_linked.pdf.
- Platforms
win· Windows- Applications
- Bluebeam Revu
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scroll- Work product
WP09· Files & folders- Conditions
licensed· Licensed pro app- Proposed difficulty
- 2 / 5
Proposed verifier
PDF parse: 64 page labels equal sheet numbers; bookmarks titles 'A-101 Floor Plan' style exact; >= 90% of callouts have link annotations to the correct page.
Proposed partial credit
35% page labels; 35% bookmarks; 30% hyperlinks.
R05.13.01.06-T12 Take off a pitched roof with area and flashing in OST
Original instruction
In On-Screen Takeoff (Windows), on sheets A-501/A-502 of project 'Warehouse W7', take off roof area for 3 roof planes with the pitch factor for 4:12 and 6:12 slopes, ridge and valley lengths, and step flashing at wall intersections. Export the takeoff summary to output/roof.xlsx.
- Platforms
win· Windows- Applications
- On-Screen Takeoff · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G3· GUI-only- Interaction primitives
path· Continuous pathgeom· Precise geometryvis· Visual judgment- Work product
WP10· Annotations- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Export quantities: sloped areas within 2% (pitch factors applied), ridge/valley within 2%, flashing within 3% of reference; conditions set with correct slope. The OST project database is not accessible outside the app and the VM has no scripting runtime, so takeoffs must be drawn in OST.
Proposed partial credit
45% sloped areas; 25% ridge/valley; 20% flashing; 10% export.
R05.13.01.06-T13 Take off drywall partitions by wall type with assemblies
Original instruction
In STACK (web takeoff, trial), on sheets A-101/A-102 take off partitions by wall type tags W1-W4 from the legend in input/partition_types.pdf, with assemblies computing studs at 16" o.c., board on both sides (W3 has 2 layers one side), and tape/mud. Wall height 10 ft. Export the material report to output/drywall.xlsx.
- Platforms
web· Any OS (web)- Applications
- STACK
- Interfaces
brow· Browser- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
path· Continuous pathvis· Visual judgmentclick· Point & click- Work product
WP10· Annotations- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Linear feet per wall type within 3% of reference; material quantities (studs, board sheets, tape) within 3% derived from correct assembly logic. STACK API keys are withheld from the agent, so takeoffs must be drawn in the web viewer.
Proposed partial credit
45% linear feet by type; 35% material quantities; 20% assemblies configured.
R05.13.01.06-T14 Pin punch items on the drawing from walk-through photos
Original instruction
In Procore (sandbox), create 12 punch items from input/walkthrough/ (photos named by room with notes). Place each item's pin on drawing A-101 in the correct room location, set trade assignee per input/trades.csv, priority, due date and attach its photo.
- Platforms
web· Any OS (web)- Applications
- Procore
- Interfaces
brow· Browser- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
drag· Drag & dropvis· Visual judgmentclick· Point & clicktype· Text & forms- Work product
WP10· Annotations- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Punch items: pin coordinates inside the correct room polygon on A-101; assignee, priority, due date and photo attachment exact. Procore API OAuth credentials are withheld from the agent, so pins must be placed on the drawing in the web UI.
Proposed partial credit
40% pins in correct rooms; 30% assignees; 15% priority/due; 15% photos.
R05.13.01.06-T15 Update field tasks on an iPad plan viewer
Original instruction
On an iPad simulator with Fieldwire (sandbox project), open plan L2 and update the 8 tasks listed in input/field_update.txt: move 2 mispinned tasks to the correct rooms by dragging, mark 4 complete with a photo from the library, and add 2 new tasks with categories and assignees at the locations described.
- Platforms
ios· iOS/iPadOS- Applications
- Fieldwire
- Interfaces
mob· Mobile touch- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & dropvis· Visual judgment- Work product
WP10· Annotations- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Fieldwire sandbox: task pin positions inside target rooms; statuses and photos; new tasks with category, assignee and location. Fieldwire API tokens are withheld from the agent, so pins and tasks must be edited in the iPad app.
Proposed partial credit
30% pins moved; 35% completions with photos; 35% new tasks.
R05.13.01.06-T16 Control: compute partition length from vector geometry by script
Original instruction
Control pair of T01. input/plans_rev3_vector.pdf is the same sheet with vector geometry on layer 'A-WALL-PRTN'. Write a script that extracts that layer's line segments, applies the 1/8" scale, merges collinear overlaps, and writes the total length to output/partitions.txt.
- Platforms
lin· Linuxmac· macOSwin· Windows- Applications
- Python · PyMuPDF
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP12· Answer/decision- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Total within 2% of 412 ft; script deterministic.
Proposed partial credit
70% total within 2 pct of 412 ft; 30% script reruns.
R05.13.01.06-T17 Count light fixtures by type with symbol search
Original instruction
In Bluebeam Revu, on lighting plans E-301..E-304 count fixture types F1-F4 using VisualSearch from the legend symbols, then review results to remove false matches (similar exit signs and smoke detectors) and add missed rotated symbols. Save count markups per type and export output/fixtures.csv.
- Platforms
win· Windows- Applications
- Bluebeam Revu
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
vis· Visual judgmentclick· Point & clicknav· Pan/zoom/scroll- Work product
WP10· Annotations- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Count per type vs reference (exact full credit, linear to 5% error); count markup positions precision/recall >= 0.95; CSV totals. Revu scripts and batch automation are disabled, so markups must be made in the Revu UI.
Proposed partial credit
60% counts per type; 30% precision/recall; 10% CSV.
R05.13.01.06-T18 Load the budget and reconcile commitments in Procore
Original instruction
In Procore (sandbox), import the original budget from input/estimate_export.xlsx into cost codes, add budget modifications for the 3 approved transfers in input/transfers.pdf, then compare committed costs from the 6 subcontracts and write output/overruns.csv listing cost codes where commitments exceed revised budget.
- Platforms
web· Any OS (web)win· Windowsmac· macOS- Applications
- Procore · Excel
- Interfaces
brow· Browserdesk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & formsmulti· Cross-app- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Budget line items per cost code equal estimate; modifications exact; revised budget totals; overruns.csv equals key.
Proposed partial credit
40% budget import; 25% modifications; 35% overrun list.
R05.13.01.06-T19 Estimate a tenant fit-out from a full plan set
Original instruction
In PlanSwift, produce the bid estimate for the 9,800 sf tenant improvement in input/ti_set.pdf (14 sheets): partitions by type, doors and hardware counts, ACT and gypsum ceilings, flooring by finish, paint areas, with unit costs from input/unit_costs.xlsx and 8% general conditions. Export output/ti_estimate.xlsx; total within 3% of reference.
- Platforms
win· Windows- Applications
- PlanSwift · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
path· Continuous pathvis· Visual judgmentclick· Point & clickmulti· Cross-app- Work product
WP02· Records/table- Conditions
licensed· Licensed pro applong· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Quantities per scope item within 3% of reference; counts exact; pricing applied; total within 3%. Only the scanned sheets are provided (no CAD). PlanSwift's plugin/scripting SDK is disabled and the VM has no other runtime, so takeoff items must be created in the PlanSwift UI.
Proposed partial credit
35% partitions and doors; 35% ceilings and flooring; 10% paint; 20% total within 3 pct.
R05.13.01.06-T20 Level subcontractor bids and award the commitment
Original instruction
For the drywall package, 4 bids are in input/bids/ (PDFs, inconsistent scopes). Level them in Excel against input/scope_sheet.xlsx, pricing scope gaps with input/plug_values.xlsx, pick the lowest compliant bidder, then create the subcontract commitment in Procore (sandbox) with the schedule of values from the winning bid and the exclusions noted.
- Platforms
web· Any OS (web)win· Windowsmac· macOS- Applications
- Procore · Excel
- Interfaces
brow· Browserdesk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
vis· Visual judgmenttype· Text & formsmulti· Cross-app- Work product
WP02· Records/table- Conditions
long· Long-horizonmultiapp· Multi-app- Proposed difficulty
- 5 / 5
Proposed verifier
Leveling sheet: adjusted totals per bidder within 0.5% of key; winner equals key; Procore commitment vendor, SOV lines and amounts, exclusions text exact.
Proposed partial credit
40% leveled totals; 20% correct winner; 30% commitment SOV; 10% exclusions.
Mapped research references
Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.
In this workflow family
- R05.13.01.01 Hotel & restaurant operations (PMS, POS)
- R05.13.01.02 Retail store & point-of-sale operations
- R05.13.01.03 Real estate & property management
- R05.13.01.04 Insurance claims & underwriting
- R05.13.01.05 Banking & lending operations
- R05.13.01.07 Facility & building automation operations
- R05.13.01.08 Service appointment scheduling (provider side)