Introducing the AI Operations Agent: Why We Built It and How to Use It Well

What's new

The AI Operations Agent is a conversational assistant built into Service Geeni. Users type a question the way they would ask a colleague, and the agent returns the answer as a chart or table with a short written explanation, in a slide-out panel inside the screen they're already working in.

It draws on the connected records in Service Geeni, including jobs, customers, sites, invoices and equipment. It also follows your existing user permissions, so nobody receives an answer built from data they couldn't already see. The agent sits within the Reporting & Business Intelligence module and is designed for one-off questions, while dashboards and scheduled reports carry on handling the figures you track every week. For a walkthrough of the interface and features, see how the AI Operations Agent works.

Why we built it

Most of the questions a service manager asks in a week can already be answered from their system. Which customer had the most breakdowns last month? Why hasn't that batch of completed jobs been invoiced? The information is there. Time is often wasted working out which report holds it and how to combine two exports, and in many businesses that knowledge sits with a small number of people.

Those people become a queue. A contract manager preparing for a customer review emails the planner, the planner gets to it on Thursday, and by then the meeting has happened without the figures. Questions also tend to arrive when something has changed, such as a spike in callouts on one site, which is exactly when a few days' wait costs the most.

There was a second reason. People are already using AI at work, whether or not their employer has chosen a tool for them. An Office for National Statistics analysis published in July 2026 found that 55% of employees reported using AI for work or education, compared with 35% of businesses with 10 or more employees reporting use of at least one AI technology. The ONS suggests part of that gap may reflect employees using AI tools independently in their day-to-day work.

In a service business, the obvious risk is someone exporting job or invoice data into a general-purpose chatbot to get a quick answer. An agent that works inside Service Geeni, under the permissions you've already set, gives people a faster route to the answer without that workaround.

Where it fits in Service Geeni's approach to AI

The Operations Agent joins the AI Scheduling Agent, which builds optimised engineer schedules and, through real-time schedule optimisation, replans the rest of the day when urgent jobs or overruns knock it off course.

The same principle runs through both, and it shapes how we think about AI in field service management more generally. The software handles the searching and the arithmetic, and people make the decisions. The scheduling agent proposes a plan that a dispatcher reviews before anything moves. The Operations Agent will explain how it reached a figure when asked, so a manager can check the working before using it to escalate a job or support a conversation with a customer.

That approach reflects what businesses say worries them about AI. In the Department for Science, Innovation and Technology's AI Adoption Research, published in January 2026, data security and the accuracy of AI outputs were the most commonly mentioned challenges in deploying AI safely. Two thirds of businesses already using AI said they applied significant human checking to what it produced. 

Which questions should you ask first?

Start with the questions you currently wait on someone else to answer. Exactly what you can ask depends on what your team records against jobs and assets, but the examples below show the kind of question that works well in the industries we support.

Material handling

For material handling businesses running customer contracts and hire fleets:

• Which trucks on our Midlands contracts have had more than three breakdown jobs this quarter?
• Which hire units have needed the most repair work between hires this year?

Ground support equipment

For GSE operations working across several stands or terminals:

• Which baggage tractors have had the most reactive jobs since the start of the summer schedule?
• How many planned services are overdue across the fleet, broken down by site?

Performance figures that arrive weeks after the event are one of the common causes of GSE downtime we've written about before. Being able to ask on a Tuesday morning closes some of that gap.

Catering and medical equipment

For service teams looking after equipment spread across many customer sites:

• Which customer sites have had repeat callouts on the same combi oven in the last 90 days?
• Which devices are due a planned service in the next 30 days and don't yet have a job booked?

Two habits help whatever your sector. Ask follow-up questions in the same conversation rather than starting again, because each follow-up stays tied to the same jobs, assets and customers as the original answer. And when a number surprises you, ask the agent how it was worked out before you act on it.

How do you get answers you can rely on?

Reliable answers come from complete records. The agent can only work with what's recorded in Service Geeni, so if engineers close jobs without notes or book parts to the wrong job, the answer will reflect that. Anything kept in a separate spreadsheet won't appear in its answers at all.

That's a common position across the sector. In the same Salesforce research, only 16% of organisations said their field and back-office technology sat on a single platform, and 52% still used spreadsheets for asset data.

A few practical steps make a noticeable difference:

• Agree what engineers must record before closing a job. Fault cause and parts used are worth prioritising, because both feed straight into questions about repeat callouts and cost.
• Keep asset records attached to the right customer and site. This matters most after equipment moves between sites or comes back off hire, because questions about repeat callouts and downtime depend on an accurate asset history.
• Use the agent to find the gaps. Asking which completed jobs have no parts recorded, or which assets have no service history, is often a quick way to see where records are thin.
• Turn recurring questions into reports. If the same question comes up every week, it belongs on a dashboard or KPI, where everyone sees the same figure calculated the same way.

Getting started

If you're an existing Service Geeni customer, speak to your account manager about getting access for your team. Because the agent follows the permissions you already have in place, it's worth reviewing user roles first, so each person can see the data their job needs and nothing they shouldn't.

If you're new to Service Geeni, book a demo to see the AI Operations Agent working alongside the rest of the platform.