By Marc Mazure

AI is now an active subject in the landscaping conversation. In July, Pro Landscaper launched a technology initiative that includes categories for business-management software, estimating technology, field-management software and AI solutions. FutureScape’s 2026 programme likewise includes sessions specifically focused on AI in landscape businesses and choosing technology for the right return.

That matters. But there is a useful distinction to make before the conversation goes any further: using AI for an individual task is not the same as being operationally ready to use AI across a business.

A business can use AI to improve the wording of an email, create a first draft of a quote, summarise notes or help with an isolated administrative task. Those are practical uses. They may save time. And they do not require an AI tool to understand how the company actually runs.

Operational AI is a different proposition. It is the idea of using technology to help someone understand a live part of the operation: which jobs may be running below expected margin; where a new maintenance contract could fit; which teams have the relevant capacity; or which projects need management attention. The examples are possibilities, not a promise about what any particular system can do today. But they reveal an important prerequisite: an AI needs the relevant, authorised context before it can give a useful answer.

The software problem nobody intended to create

In my conversations with landscaping businesses, I regularly see a mix of systems that have each been adopted for good reasons. There may be accounting software, an HR platform, estimating software, spreadsheets, folders on local computers, Dropbox or other file storage, WhatsApp, a CRM, a task-management tool and a separate scheduling process.

That is not evidence that the business is badly run. It is usually evidence of a business solving real problems as it grows.

Someone needs to manage payroll and leave, so HR software is introduced. Someone needs to prepare estimates quickly and consistently, so a specialist estimating tool is added. A team needs to coordinate work, so a scheduling platform or spreadsheet develops around the way that team works. Site photos, messages and updates need to move quickly, so people use WhatsApp. Finance needs a clear accounting record, so that sits in a dedicated accounting package.

Each decision can be entirely rational. The difficulty is not necessarily within the tools. It can be in the space between them.

Information about a vehicle’s MOT date may be in a spreadsheet. Project information may be somewhere else. Staff availability may sit inside HR software. Schedules may be managed in a different system. Important site detail may be held in a conversation. Over time, people can become the bridge between those places: checking, exporting, uploading, copying, reconciling and remembering.

That is why some business owners ask for “one system.” Often, they are not saying that every specialist tool is poor. They are describing the pressure of having to search in several places to understand what is happening and then manually connect the information that should already be related.

Digitised is not necessarily connected

It is possible for a company to digitise much of its operation without creating a connected operational picture. That is an analytical distinction used in this article, not a claim that it is established industry terminology.

Consider a simple HR and scheduling example. A business may use an HR system to record holiday, sickness and perhaps time. Its schedule may exist elsewhere. Both processes are digital, and both systems may work well. But if a scheduler still has to check one place and manually update or adjust another, the operational connection depends on a person.

The same can happen from estimate to project to invoice. Customer information may be managed in a spreadsheet or CRM. A dedicated estimating system creates the quote. Details are exported and moved into a project or task platform. Further information is recreated or uploaded. Invoicing is completed in the accounting system because that is where the finance process lives.

Again, none of this criticises the software. The issue is the manual movement of information between stages. It affects how quickly a person can see the full picture, and it may affect whether an AI tool can work reliably with the information needed for a particular operational task.

Connected does not have to mean that every record is moved into one place. It can mean that the business has deliberately decided where each important record belongs, how it is updated, and how the right information is made available to the right person at the right point in a workflow. A payroll system can remain the right home for payroll. A specialist estimating tool can remain the right home for estimating. The operational question is whether a confirmed holiday, accepted quote, revised job date or overdue invoice can be understood where it needs to be understood, without a member of staff having to reconstruct the story from scratch.

That is also why data quality matters. If a job is called one thing in the estimate, another in the schedule and something else in finance, joining the information together becomes unreliable. If a site instruction exists only in a chat thread, it is difficult for a new team member—or an approved AI assistant—to find, assess or act on it. Improving this does not require perfect data. It requires attention to the information that is important to a recurring operational decision.

Task-level AI and operational AI are not the same thing

Task-level AI can be useful without deep context. If a manager wants help drafting a client email, the AI only needs the instruction and the content the manager chooses to share. If an estimator wants a better first draft of a proposal, the tool may only need the relevant project brief and an approved template.

Operational AI would need more. Imagine asking:

“Where can we fit this new maintenance contract into next week’s schedule?”

A responsible and useful answer could require existing jobs, locations, estimated durations, staff availability, holiday and sickness, relevant skills, customer commitments, contract requirements, recurring visits, travel or geographical clustering, and the live workload already committed. It does not need every record the business has. It needs the relevant authorised context for that decision.

If those pieces of information are spread across disconnected systems, spreadsheets, conversations and people’s memory, an AI does not automatically acquire the complete view simply because the company has started using AI. It may return something that sounds plausible while missing a change in leave, a key site constraint, a promised visit or a team member’s availability.

That is a reason to be thoughtful, not fearful. The objective is not to hand decisions to an AI without review. It is to understand what information a person and an AI would need in order to make a better first recommendation, surface an exception or prepare a useful starting point for someone to check.

There is a governance point here as well. An AI should not automatically be given unrestricted access to every staff, customer or financial record simply because it may be useful. A business needs to decide what information is relevant to the task, who is authorised to see it, which source is trusted, and where a human must remain responsible for the final decision. In the scheduling example, an AI might suggest options. A manager should still judge whether the work is commercially sensible, whether a customer commitment has nuances that are not captured in a system, and whether the proposed plan is right for the team.

That approach makes operational AI more practical. It starts with a bounded use case, clear permissions and a person who can test whether the recommendation reflects reality. It is less about creating a perfectly automated business and more about improving the quality and speed of everyday operational judgement.

The distinction also reflects a wider point made in the UK Government’s 2026 AI Adoption Plan for Professional and Business Services. The report says AI use often helps people complete existing work more quickly, but, without changes to workflows, decision-making and organisational design, those benefits remain localised rather than becoming firm-wide productivity gains.3 The report is not a survey of landscaping businesses, so its figures should not be applied to the sector. Its wider lesson is still useful: access to an AI tool is not the same as an operating model that can use AI well.

The quieter work that makes future AI more useful

It is tempting to start with the question, “Which AI tool should we use?” For many landscaping businesses, a better first question may be: “What decision would we like better help with, and what information would that require?”

Take the scheduling example. Work backwards from the decision. Where do job dates live? Who owns the record of staff availability? Are skills recorded in a usable way? Is a recurring contract stored in the same place as the current schedule? Are site notes accessible and reliable? Does anyone still need to re-key information between systems before a manager can act?

This is not a case for replacing every tool with one large system. A connected operation can use specialist software. The important questions are whether the information is clear, whether the right system owns it, whether critical systems share it appropriately, and whether people can retrieve what they need without relying on memory or manual transfers.

A practical starting framework could look like this:

This work has value even before AI is involved. It reduces duplicate entry, makes handovers clearer and helps people find information faster. It can also make future technology choices more disciplined, because a business knows what information it needs to connect and why.

A small pilot can make the exercise real. Rather than trying to make an entire operation “AI ready”, choose one decision that happens every week: allocating a crew, reviewing a job that is drifting beyond its budget, or preparing for a recurring contract. Map the information that decision genuinely needs. Note who updates it and whether it is current. Then identify one improvement that would make the decision easier for a manager today. If that foundation is not useful to a person, it is unlikely to be useful to an AI.

A more grounded view of AI readiness

The landscaping businesses best placed to make operational use of AI may not simply be those experimenting with the most tools. They may be the ones that have created a clearer view of the information that runs the business: where it lives, who owns it, how it is updated, who can access it and how it moves from one operational stage to the next.

That does not require a grand transformation programme. It may start with a single recurring frustration: holiday information that does not reach the schedule, estimate details that have to be recreated in a job system, site updates trapped in individual conversations, or job-cost information that arrives too late to be useful. Those are not merely administrative inconveniences. They are clues about where the operational picture is breaking apart.

CQ’s approach to landscaping business management software is based on making relevant operational information easier to connect and use. The point of this article, however, is broader than any one platform: a business should first understand the information and decisions that matter to it. If you would like to explore that connected-operations approach in the context of your own business, you can book a CQ demonstration.

The promise of AI will not be fulfilled by asking a tool to do more while the context it needs remains scattered. The more durable opportunity is to build an operation where the right people—and, where appropriate, the right AI support—can work from a clearer, connected and properly governed view of what is happening.