UK contractor Bojen Electrical uploads tender drawings and builds its takeoff with automatic symbol recognition, keeping a permanent, editable count list for each job. The estimator says that working through the drawings this way also gives him a clearer picture of how the job will actually run. He priced a roughly £2.5 million lighting refurbishment of an Oxford research building entirely in the tool, splitting each floor into zones.
High Ground (about 150 employees across NJ, NH and ME) started with AI only on overflow and after-hours calls in place of voicemail. It then moved AI to the front line one day a week, then two, then all five, with warm transfers to CSRs for escalations. The company reports AI now handles 75% of inbound calls, the abandon rate fell from 10% to 2%, AI books 81% of bookable calls, $250K-$300K a year in CSR costs were avoided without layoffs, and outbound SMS campaigns bring in $100K-$250K a month.
Hatch, ServiceTitanVendor case studyHatch · Dec 2025
Intermountain Electric's estimators run drawings through AI for first-pass lighting, power and receptacle counts, branch routing, panel and lighting schedules, and drawing stitching, then review and correct it before anything goes into the estimating system. The case study says one high-volume takeoff that used to take several 15-hour days was done in a few hours, and that the AI gets branch measurements about 95% of the way before human QA.
Family-owned WTC Electric in Charlotte, NC faced rising bid volume, driven partly by data-center demand, and couldn't find another estimator. It moved takeoffs to AI device detection and automated branch routing, with the vendor's human concierge team handling peaks and poor-quality drawing sets. Drawer AI reports a 70% reduction in takeoff time on large projects and device detection accuracy above 95%.
Florida's Del-Air (eight locations, 220+ technicians) put AI on dispatch first. Within three months it centralized dispatch at one office and went from 22 dispatchers to 10, with about 90% of call assignments now made automatically. It then made AI the first touch on all inbound calls: in May the AI handled 16,000 calls and resolved 74% of them without a person.
Brothers had call abandon rates near 30%, and callbacks could take 10+ hours. The team set up ServiceTitan's AI Virtual Agent around its real capacity, zones, technician skill sets and job types so it only books what the business can deliver. ServiceTitan reports 600+ jobs booked by the agent, $1.33M in attributed revenue and a 67% drop in the abandon rate (to about 5%). "Since November, we've recovered over a million dollars worth of revenue," says lead systems administrator Tyler Frydenlund.
At RL Electric, the person who prepares proposals used to count symbols by hand with PDF markup tools across dozens of sheets per bid, where one missed symbol could throw off the price. The team switched to AI symbol recognition and counting directly on the electrical plans, then verifies and adjusts the counts. Less repetitive counting left more time for bid strategy.
German electrical contractor Rudolf Fritz (about 650 employees) uses automated quantity takeoff from PDF plans for bids. In the first five months after its March 2026 rollout, it processed 364 plans across 18 projects, detecting 71,073 elements and measuring 85 km of cable tray. Takeoffs now take hours instead of weeks, and the team says it can price more inquiries instead of passing on projects for lack of capacity.