The field service & AI dispatch glossary.

▸ The dispatch loop

AI field-service dispatch

The practice of running the field-service dispatch loop with AI agents: an agent reads an incoming work order, drafts a price, ranks and routes the right contractor, and tracks the job to close, while people supervise the judgment calls. It contrasts with legacy field service management software, where a human dispatcher does the reading, quoting, and routing by hand and the software only stores the record.What is AI field-service dispatch?The Dispatch Engine

Dispatch loop

The repeating cycle every field-service job moves through: intake, quoting, routing, and close. AI field-service dispatch places agents on the routine work at each stage of the loop, and escalates the exceptions to a person.What is AI field-service dispatch?

Dispatch Engine

STEADYWRK's agentic dispatch system. It takes a work order in, drafts a not-to-exceed quote, ranks and routes contractors, and reconciles the outcome — exposed as callable tools over a published API rather than a dashboard a person clicks through. It runs the four-stage dispatch loop with a human on the exceptions.The Dispatch Engine

Intake

The first stage of the dispatch loop: reading an unstructured request — an email, a PDF, a webhook, an API call — into a structured job with trade, location, urgency, scope, and contact. Reading messy input into clean fields is the hard part; legacy software waits for a person to key every field.

Not-to-exceed quote (NTE)

A price ceiling agreed before work begins. The contractor may bill up to the not-to-exceed amount without a new approval, and anything above it requires sign-off. In the dispatch loop an agent drafts the NTE from the scope, and a person reviews and owns the number.

Routing

Ranking candidate contractors by trade fit, proximity, completion history, and price alignment, then assigning the job. Agentic routing runs at any hour, so throughput is no longer capped by one dispatcher working tickets in sequence.

Close and reconciliation

The final stage of the dispatch loop: the job is chased to completion, reconciled against the quote, and recorded. Every decision along the way is written to the audit log, so the job leaves the loop as a closed, inspectable record.

Work order

A request for field-service work — a repair, an inspection, a maintenance task — carrying a trade, a location, a scope, and an urgency. The work order is the unit that moves through the dispatch loop from intake to close.

Dispatch latency

The time between a work order arriving and a contractor being routed to it. Lower latency means less dead air between the request and the person who does the work, which is why latency is a headline measure of a dispatch operation.

▸ Autonomy and trust

Agentic control plane

The coordination layer where AI agents run operational work end to end — reading inputs, deciding, and acting through callable tools — while humans supervise by exception. It is the system that runs the work, not just a screen that displays it.What is an agentic control plane?The control plane

AI operator

A person who supervises AI agents doing operational work. They set the goals, handle the exceptions the system escalates, and own the judgment calls, rather than performing every step by hand. The role is supervision and accountability, not manual execution.What is an AI operator?

Human-in-the-loop

A design where automated decisions can run on their own up to a confidence threshold, and anything below it is escalated to a person. The goal is leverage with a brake — autonomy where it is safe, human judgment where it is not.

Confidence gate (policy gate)

The rule that governs autonomy. High-confidence, low-risk decisions proceed automatically, while low-confidence or high-risk decisions are escalated to a human for review. The gate is where an operator decides how much of the loop runs on its own.

Escalation

Routing a decision the system is not confident enough to make on its own to a person. Clear, fast escalation is what keeps an autonomous loop accountable as volume grows — uncertainty becomes a visible signal up the stack instead of a silent guess.

Append-only audit log

A tamper-evident record where every decision is written once and never altered, so any routing or quoting decision can be inspected after the fact. In an agentic system, trust rests on the receipt, not the claim.

Eval

A repeatable test of how a system actually performs on real tasks, published and refreshed on a rolling basis. Evals let a buyer judge an agentic system on measured behavior instead of a sales-deck figure.Public evals

Human override

A person stepping in to change or reverse an automated decision. A low and falling override rate over time is a sign the agents are learning the work; a sudden spike is a signal to investigate the inputs or the policy.

▸ Field service and FM

Facility management (FM)

The discipline of keeping buildings and their systems — HVAC, electrical, plumbing, life-safety — running, through preventive maintenance and on-demand repair across a portfolio of sites. FM is the demand side that generates the work orders a dispatch operation routes.AI facility management

Integrated facility management (IFM)

A model where one provider coordinates the full range of facility services across a portfolio under a single accountable relationship, rather than the client managing many separate trade vendors. IFM concentrates coordination — exactly the work an agentic control plane is built to run.

Field service management (FSM) software

The system-of-record category a human dispatcher operates. It stores work orders, shows a schedule, and tracks technicians, but a person still reads each ticket, types the quote, and picks the contractor. AI field-service dispatch moves that loop off the dispatcher and onto agents.AI dispatch software

CMMS

A computerized maintenance management system: software that tracks assets, schedules preventive maintenance, and stores maintenance history for a facility or portfolio. A CMMS is a record of the asset; a dispatch system is the engine that acts on the work the CMMS surfaces.

Service-level agreement (SLA)

A commitment on how a service will be delivered — for example a response or completion window for a category of work order. SLAs set the expectation a dispatch operation is measured against, and they are the lines a confidence gate is tuned to protect.

Trade

A skilled field-service discipline such as HVAC, plumbing, electrical, carpentry, or general maintenance. The trade on a work order drives which contractors are eligible to be routed to it, and which market wage baseline applies.Labor rate explorer

Preventive maintenance

Scheduled, recurring service done to keep equipment running and prevent failures, as opposed to reactive repair after something breaks. Preventive maintenance generates a predictable stream of work orders, which is the kind of volume agentic routing handles well.

Scorecard

A structured rating of a contractor or applicant on the factors that predict good outcomes — trade skill, reliability, completion history — used to screen people in and to route work fairly. The scorecard turns a hiring or routing judgment into something repeatable.Talent

▸ Platform and integration

Model Context Protocol (MCP)

An open standard that lets AI assistants connect to external tools and data through a common interface. Exposing a dispatch system over MCP lets other software and agents drive it directly, instead of a person clicking through a screen.IntegrationsDevelopers

API-first (callable tools)

A design where the core operations are exposed as a published API and callable tools that other software can drive, rather than being locked inside a single screen a person clicks through. API-first is what makes a dispatch system machine-discoverable and composable.

Talent fabric

STEADYWRK's placement layer. Scorecard-screened, bilingual field-service talent is matched into roles, so demand for work and supply of qualified people meet inside one operation rather than across disconnected job boards.TalentFor employers

Nearshore

Staffing operational and support roles from a nearby region in a compatible time zone, combining a cost advantage with real-time collaboration. STEADYWRK runs nearshore operations from Aqaba, Jordan, serving both Western and regional clients.Nearshore Jordan

Bilingual operations

Running field and support operations natively in both Arabic and English, so the same team can serve Middle East and Western clients without a translation seam. Bilingual capability is an operational property, not an add-on.

Sovereign AI

AI infrastructure and operations a company owns and controls end to end — its models, data, and orchestration — rather than renting capability it cannot inspect or govern. Sovereignty is about who holds the controls and the data, not only who runs the servers.Ghazal

▸ Region and compliance

Nitaqat

Saudi Arabia's workforce-nationalization program. It rates private employers on their share of Saudi nationals and ties hiring privileges to that rating, so it shapes how companies operating in the Kingdom build a field workforce.Nitaqat

Ajeer

A Saudi labor program that governs the compliant use of contracted and transferred workers between employers, keeping sponsorship and authorization correct when one firm performs work under another. It is part of the regional compliance landscape FM operators navigate.Ajeer

ASEZ (Aqaba Special Economic Zone)

Jordan's special economic zone around Aqaba, with its own regulatory and customs regime designed to attract investment and operations. It is STEADYWRK's home base — the place the operation is built and run from.Aqaba