BRENNEN

Industries / Technology / Software and SaaS

Field service software

Field service software is what companies that send technicians to customers use to book visits, decide who goes where and when, put each job on the technician's phone, keep the right parts on hand, price and bill the work, keep the customer informed, and help the technician from a distance.

1983

Motorola built a radio data network that let IBM's service technicians use Motorola KDT800 portable two-way data radios to reach IBM's computers. IBM later used the network to dispatch its technicians and coordinate parts orders.

motorolasolutions.compenbasedcomputing.com

30

Milestones achieved across 6 domains.

10

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

In 1983 Motorola built a radio network so that IBM's service technicians could reach IBM's service databases from handheld terminals. The same year Otis opened OTISLINE, a centre that took every service call around the clock and sent mechanics out from there. Operations researchers worked out the planning: test problems for routing visits inside time windows (1987), and a scheduler for 20,000 engineers that BT said was saving it $150 million a year by 1998.

Where it stands

Three analyst firms put the market for field service management software at $5.5 billion to $6.6 billion in 2025, and ServiceTitan alone says its customers ran $62.0 billion of jobs through its software in a year. Technicians now carry ordinary phones, and the headsets once sold for remote help are being withdrawn. Large vendors sell AI agents that book, schedule and write up jobs, with figures that are their own; the regular outside measures of appointment keeping come from telecom and energy regulators, and Australia's found that NBN Co met about 56% of its appointment times.

What is open

The open jobs below are public tests and datasets: a bad-day test for AI dispatchers, voice notes recorded in loud plant rooms, data plates read from photos, parts predicted from fault reports, AI quotes checked against real ones, an on-time score any firm can compute, and AI answers checked against service manuals. Each can be started this month with open-source solvers, current speech and vision models, and records the builder has permission to use.

Domains

AchievedWatch forOpen

Advances to build with

1 research note

Open in field service software

3 of 12 open deliverables

Open · Scheduling and dispatch

A bad-day test for dispatchers

A public benchmark that replays a working day for a team of technicians and adds what goes wrong: jobs that run over, cancellations, a technician off sick, an urgent job arriving at noon. It scores AI agents built on language models against standard optimization solvers on travel time, late arrivals and jobs left undone. A second condition compares one agent planning for the whole team with one agent per technician.

Built with what exists

Google OR-Tools, Timefold, VROOM or PyVRP, open-source solvers that handle time windows and skills, as baselines; the real-world TDC Net test set from the Technical University of Denmark (720 technicians, 2,677 tasks, 31 skills, published on Zenodo) and test sets derived from Solomon's; a coding agent such as Claude Code to run the language-model side; the Note "Many agents, one budget" for the one-agent-or-many question.

Credit · unclaimed

Open · Work orders on mobile

Voice notes in loud rooms

A public test of how well speech-to-text tools and AI form-fillers turn a technician's spoken notes into a correct work order when the notes are recorded where technicians work: beside running air-conditioning units, in boiler rooms, in vans with the engine on. It scores word errors and, separately, whether each field of the work order (fault found, parts used, readings taken) is right.

Built with what exists

ElevenLabs Scribe v3 (see the Note "Fewer word errors") and OpenAI's open-source Whisper for transcription; a small model such as Meta's Muse Ember to fill the work order fields (see the Note "The jobs agents repeat"); phones and a sound level meter app; scripted notes read aloud by technicians.

Credit · unclaimed

Open · Parts and inventory

Parts prediction, tested in public

An anonymized public dataset of field service jobs, from several firms and trades, that links each fault description and equipment model to the parts actually used. With it, a baseline model that ranks the likely parts, and a scoreboard for others to beat.

Built with what exists

exports from field service software, used with the firms' permission; Python to strip personal data; retrieval of similar past jobs, as in the Italian system; small language models fine-tuned on one graphics chip (see the Note "One GPU to fine-tune").

Credit · unclaimed

Where it sits