The Quote That Becomes a Contract on Your Landscaping Site

A homeowner messages your landscaping site: "what would it cost to redo the whole backyard with a patio and new sod, and is the sod guaranteed to take?" The website assistant, eager to win the work, throws out a price and promises the sod will thrive. No one walked the yard. No one saw the drainage, the slope, or the soil. By the time your crew shows up, there is a number and a guarantee on the record that the business never agreed to.

Landscaping looks like a low-stakes service to automate, and that is exactly why the exposure sneaks up. A general-purpose chatbot treats a pricing question and a "will it last" question as things to answer. In a trade that runs on site-specific estimates, applied chemicals, and built structures, those answers carry weight the company has to honor.

A Firm Number Sight-Unseen Becomes a Term

When a homeowner asks what a project will cost, the honest answer depends on the site itself, on the materials, on access to the yard, on the full scope. A website knows none of that. A generic bot answers anyway, because answering is its job. The figure it produces is not a casual estimate to the person reading it. A price a customer relies on can become a term of the deal, and the gap between the chatbot’s number and the real bid is where the dispute starts. The same is true of a promised completion date the crew never committed to.

"Guaranteed to Take" Is an Express Warranty

The guarantee is its own trap. When the bot says the sod will thrive, the trees will survive, or the patio will never crack, it is making an affirmation about the work. Under the rule that an affirmation or promise about goods creates an express warranty, statements like that can bind the business to a result, and a customer is entitled to rely on them. The company did not decide to warrant the sod against a bad season. A website widget did, in one reassuring sentence.

Misrepresenting what a service will deliver also sits within the FTC’s authority over unfair and deceptive practices, so an inflated promise that draws a complaint is not only a contract problem.

The Chemicals Have Their Own Rulebook

Landscaping that involves herbicides, pesticides, or certain fertilizers picks up a second regime. Under federal pesticide law, using a product inconsistent with its label is a violation, and the label, not a chatbot, dictates rates and precautions. A bot that tells a homeowner what to put down on the weeds, or that the application is "safe to walk on right away," is giving guidance the label controls. Add the state licensing that applies to applying chemicals and to certain hardscape and irrigation work, and the website is improvising across several regulated lines at once.

"Will Not" Is a Suggestion. "Cannot" Is an Architecture.

Here is what trips companies up. A homeowner who wants a price does not ask once and accept "we’ll send someone to estimate." They push. "Just ballpark it, what does a yard like mine usually run?" A chatbot steered only by a prompt eventually answers the version of the question it was not specifically warned about, because being helpful is its default and a prompt is only a polite request to suppress that default.

That is the gap between a tool that is told not to quote and one that is built so it cannot. A disclaimer bolted onto a system whose whole purpose is to give confident answers does not change what it does when a motivated customer leans on it. It just leaves the company holding the number the bot invented.

Who Owns the Answer

Strip away the software and the exposure is familiar. An employee who quoted a firm price without seeing the yard and guaranteed the sod would create a headache the owner has to clean up. When an unsupervised chatbot does it on the website, the liability does not disappear. It lands on the business. Now there is a written record, no estimator was in the loop, and a guarantee the customer can point to.

The companies that get burned are not the ones that modernized their sites. They are the ones that dropped in a generic chatbot, assumed helpful and safe were the same thing, and learned the difference in a dispute over a number nobody at the company ever approved. The fix is not to take AI off the site. It is to run an assistant that books the walkthrough, states only what the company has published, and leaves the bid and the guarantees to a person.

Frequently asked questions