What is AI dispatch software?

For service businesses, AI dispatch usually means: prioritising jobs, recommending the best technician, suggesting a route order, and drafting customer ETA messages. It’s rarely fully autonomous dispatch.

Should I let AI auto-assign jobs?

Only after you’ve proven your data is clean (job types, skills, availability) and you have an exception process. Most teams start with AI recommendations and a human approving assignments.

What data does AI dispatch need to work?

At minimum: job location/postcode, urgency/SLA, job type, technician skills, working hours, and current schedule. Without those, AI guesses—and guesses break customer trust.

AI dispatch software for field service: what works in 2026

When vendors say “AI dispatch”, most trade businesses hear: “it will magically fix scheduling”.

In reality, the wins come from four boring things:

Clean job intake (right job type, right address, right urgency).
Clear skill tags (who can do what).
A prioritisation rule everyone agrees on.
A dispatcher who approves the edge cases.

What “AI dispatch” can safely do

1) Triage: decide what matters today

AI is good at scoring jobs if you give it strict inputs.

Example scoring signals:

urgency (emergency vs routine)
customer impact (no heat, lockout, leak)
revenue impact (deposit collected, high value)
breach risk (SLA, promised window)

2) Recommend technician + slot

AI can rank options, but should not override reality:

travel time
skill match
parts availability
access windows

3) Route order suggestion

This is where AI shines: reorder the day to reduce drive time. But the dispatcher should sanity-check:

fixed appointment times
multi-day jobs
jobs requiring two people

4) Draft customer comms

The fastest value: consistent updates.

booking confirmation
“on the way” with ETA
delay message with options

What AI should not do (without strict guardrails)

Promise ETAs without checking a live schedule.
Book emergency jobs into impossible slots.
Decide pricing or discounts.
Override safety policy (“just turn it off and see”).

A simple dispatch policy (copy/paste)

Dispatch policy

Prioritise safety and customer harm first (gas smell, active leak, lockout, vulnerable customers). Then protect promised windows. Then optimise travel time. Revenue comes last.

How to implement this in Toolfy (without a rebuild)

Store the minimum dispatch data

If you want AI suggestions to be accurate, you need consistent fields:

Job type (e.g., “boiler breakdown”, “EICR”, “storm repair”)
Urgency (routine / urgent / emergency)
Skills tags (e.g., “gas-safe”, “ladder-work”, “EICR”)
Location (postcode + address)

Toolfy already gives you the schedule, job statuses, and customer records. The missing piece is often just tagging discipline.

Use AI as a “dispatch co-pilot”

Recommended first rollout:

AI reads today’s job list (export or API read).
AI produces recommendations:
- top 3 priorities
- suggested assignment per job
- route order per tech
Dispatcher approves changes in Toolfy.

When you’re ready, you can connect AI tools to Toolfy actions (create/update jobs) using ChatGPT Apps/MCP—but keep idempotency and scopes tight.

Dispatch prompts (for co-pilot mode)

Prompt: daily triage summary

You are the dispatcher for a UK trade business.
Given today's jobs and technicians, output:
1) the top 5 priorities with reasons,
2) recommended assignments with confidence (high/medium/low),
3) any schedule conflicts or missing info you need to ask about.
Do not invent details. If required info is missing, list questions instead.

Prompt: ETA message (no promises)

Write a customer update:
- confirm the appointment window
- give an ETA range (not a precise time)
- include one sentence if we are running late and what the customer can do
Tone: calm, professional, UK spelling.

When AI dispatch is worth it

2+ techs (coordination becomes real)
high emergency volume
jobs spread across a wide area
you frequently “forget” to send customer updates

If you’re solo and local, AI dispatch won’t beat a simple checklist.