Most landscaping and hardscape owners have already tried ChatGPT once. You paste a question, get a polished answer, shrug, and go back to answering the phone. That is not the same as putting AI to work in the business.
The useful applications are narrower than the hype. They reclaim hours on phones, proposals, hiring, and numbers — and they still need a human who knows the work to catch when the model is wrong.
This guide pulls from Applications of A.I. in a Landscaping Business with Jeremy Gunness of Knowledge Tree Consulting on the How to Hardscape Podcast (episode 379). Jeremy is CFO and CTO at KTC. He comes from mechanical engineering, telecom, and banking tech, co-founded KTC with roots in the Oriole Landscaping experience, and spends his time building practical systems for landscape and hardscape companies across the US and Canada.
What “AI in a Landscaping Business” Actually Means
Large language models (ChatGPT, Claude, Gemini, Copilot, and the rest) are trained on past human writing and data. They are good at producing a plausible next answer. They are not good at knowing when that answer is invented.
Jeremy’s core point: the best results happen when you stay close to work you already understand, so you can spot nonsense. The danger rises when you ask an LLM to run a whole marketing plan, design language, or financial strategy in a domain you cannot verify.
So the goal is not “run my company with AI.” The goal is one painful admin loop at a time — less hours, more output, same quality of judgment.
Watch Out for AI Washing
A lot of software is now labeled “AI” for features that barely save time. Some of the interesting capabilities are still in beta. Price is a useful filter. If an “AI” add-on costs the same as or more than the human hours it replaces, be skeptical. Michael’s example on the episode was a “first AI bookkeeper” demo that landed around seven times his actual bookkeeper cost, with the justification essentially being “because it’s AI.”
A better pattern is narrow and grounded. Jeremy pointed to Hostinger’s support chatbot as an example outside our industry: it can investigate a specific SSL problem and fix configuration with the users authorization. Clear job. Clear tools. Permissioned. That is closer to how AI should show up in a landscaping business than a vague “build me a marketing plan” button.
AI Phone Agents for Landscaping Companies
This was the most concrete build Jeremy described.
Classic phone trees are not enough. An intelligent phone agent can filter spam, identify why someone called, move hot prospects to the front of the line, answer simple FAQs (hours, directions, services) by text or email, and push a summary into your CRM (HubSpot, custom systems, or field-service software like Jobber). The same funnel can include your website contact form and a site chatbot so every inbound lead lands in one place.
Who it helps:
- Owner-operators whose personal cell is also the company line — spam, clients, employees, and complaints all hitting one phone while you are trying to run crews.
- Multi-division firms (retail yard plus landscaping) that currently force an office manager to route every call by hand. Static sod pricing can be answered one way; a patio project lead should go to landscape sales.
Design rules that protect your brand:
- Keep conversations short and simple.
- Do not loop callers through endless menus.
- Escalate frustration to a human immediately.
- Do not over-automate payments or complex orders on the agent.
- Monitor calls and tweak the agent over time. The same script will not feel the same on every call, so complexity is the enemy.
If your telecom provider already offers an AI receptionist, Jeremy’s advice was practical: try theirs first before you build something custom.
Turning Estimate Data into Proposal Drafts
Jeremy’s team built an estimation app for a tree company. Once the estimate data already exists on iPad or desktop, one button plus a prompt can draft a customer-facing proposal in roughly twenty seconds.
That is not “upload a photo and two sentences and get a full estimate. The numbers and scope still have to come from your process. AI packages the write-up. You still read it, edit it, and verify every dollar, because models will invent details when they are unsure. The target is an 80–90% draft you polish in minutes instead of writing from scratch for an hour — meaningful when you have a stack of estimates sitting in the queue.
Hiring Without Drowning in Indeed Spam
KTC has been building an HR intake system (Oriole as an early recipient) that starts from real job descriptions — they keep a library of dozens of landscape roles. AI generates a qualifying questionnaire an admin can edit. That friction alone cuts down “click to apply” spam.
Then AI can score applications against the job description (answers, resume, cover letter), draft interview invitations, polite rejections, and next-step notes. You still decide who gets a call. The win is hours back for owners who are also the HR department.
Using AI on Your P&L and Day-to-Day Numbers
As a CFO lens, Jeremy’s use case is simple: feed year-over-year P&L and balance sheet data into an LLM and ask for patterns and talking points before you sit with your accountant. It is a discussion starter, not gospel.
The same tools help with bookkeeping questions — how to handle an odd journal entry or payment — faster than a generic web search, which can reduce expensive “how do I…” accountant time. You still know what went into the bank, what debt you carry, and what cash is moving. Use that reality to sanity-check anything the model says.
Also backup spreadsheets before you let Copilot or Claude plugins rearrange files. Mutations are easy. Undoing them is not.
Everyday Office Use Without Losing Your Voice
For client email, Jeremy’s rule is write the first draft yourself. Then use Copilot (or a similar polish tool) to clean it up. Letting the model author cold from scratch often sounds robotic and costs you the voice your clients already know.
On creativity and design, stay careful. Generative tools (including video and image models like Sora) tend toward average, uncanny sameness when you ask for a client backyard mockup. Landscaping and hardscape are craft and relationship businesses. Mass “100 agents making 100k cold calls” lead spam sits badly against that culture. Use gen tools as aids. Do not hand them the design or the sales identity of the company.
How to Start This Week
- Pick one painful admin loop: phones, proposal write-ups, hiring inbox, or monthly numbers review.
- Open Claude (Jeremy’s daily favorite), ChatGPT, Gemini, or Copilot if you are on Microsoft 365. Ask better questions inside domains you can verify.
- Never skip human review on money — estimates, proposals, and financial conclusions.
- If phones are the bottleneck, define triage rules, FAQs, CRM logging, and a fast human handoff before you buy anything with “AI” in the logo.
- Price-check vendors. Automation should cost less than the human hours it replaces.
- Protect creativity and client relationships. Narrow tools beat hype wrappers.
If you want the full conversation, listen to Applications of A.I. in a Landscaping Business with Jeremy Gunness of KTC on the How to Hardscape Podcast. For consulting and Smart Growth Tour details, Knowledge Tree Consulting is at ktc.biz (Jeremy: jeremy@ktc.biz).