Lesson objective
Write behaviour objectives for the five core AI-assisted tasks in a landscape company, each tied to the record it produces and the person who owns the decision behind it.
Plain-language explanation
Why this matters now
Most of what AI does in a landscape company falls into five families of task: estimating, scheduling, closeout, identification and reporting. Each produces a record that someone relies on: a proposal a customer signs, a route a crew drives, a note that becomes an invoice, a plant or pest name that drives a decision, a report that goes to a manager or an authority. An objective written against one of those records is testable. An objective written against a topic is not.
This lesson writes the objectives. It does not teach the qualified work behind any of them. Where a task touches a pesticide decision, a supervision duty or an employment record, the objective is written so the learner routes, records or checks, and the qualified person or the employer decides.
As in the rest of this level, the method is an AI Coalition Network recommendation, not a finding from learning research.
A question to sit with
Your scheduler uses an AI tool to redraft Tuesday's routes after a rain day. Your crew leaders use one to turn voice memos into closeout notes. Your estimator uses one to draft plant lists. If you could change one thing each of them does, would you know which record to look at to see whether it changed?
What goes wrong
1. One objective tries to cover every task. "Staff will use AI responsibly" covers estimating, scheduling and closeout at once and changes none of them. Each family fails in a different way and needs its own objective.
2. The record is left out. An objective about closeout that does not mention the visit record, the time record or the invoice cannot be checked, because nobody knows where to look.
3. Identification is treated as a lookup. An objective such as "crew leaders will identify weeds with the app" treats the tool's answer as the identification. The answer is at most a starting point for a person who can confirm it.
4. Reporting objectives ignore what the law already requires the record to contain. Some records are not the company's to design. Time records have required contents; records of who was trained to use restricted use pesticides under supervision have required contents. An AI-drafted version that drops a required field is a compliance problem, not a formatting one.
5. The owner is missing. Every one of these tasks ends in a decision someone owns. An objective without the owner invites the learner to assume the tool made the decision.
The practical answer
AI Coalition Network recommendation: one objective per task family, written against its record and its owner.
| Task family | Record it produces | Example behaviour objective | Who owns the decision |
|---|---|---|---|
| Estimating | The proposal the customer signs | Before a drafted proposal leaves the office, the estimator checks every plant, quantity and price against the site notes and the availability sheet, and removes any promise the company cannot keep | The person who signs the proposal |
| Scheduling | The route and the crew assignment | Before a redrafted route is released, the scheduler confirms every skipped property is listed as skipped and every crew's hours fit the day | The scheduler or operations manager |
| Closeout | The visit note, the time record, the invoice | Before sending a drafted closeout note, the crew leader deletes any line the visit record does not support and confirms the hours match the time record | The crew leader, then the office |
| Identification | The plant, weed, pest or disease name that drives a next step | When a tool names a pest or disease, the crew leader records it as unconfirmed and routes any treatment question to the certified applicator | The certified applicator for any treatment |
| Reporting | Reports to managers, customers or authorities | Before an AI-drafted report is sent, the author checks each figure and required field against the source record | The named author and their manager |
Five objectives written this way give you five practice tasks and five places to look afterwards.
Know which records have required contents. Teach the objective with the law that shapes the record, stated precisely.
- Law / rule (federal, binds FLSA-covered employers): 29 CFR 516.2 requires, for each non-exempt employee, among other items, the hours worked each workday and the total hours each workweek. The regulation does not require any particular form of timekeeping, and it is indifferent to medium: a mobile or AI-assisted time record meets it only if it actually captures the required items and can be produced when asked. Under 29 CFR 516.7, records must be produced within 72 hours of notice from the Wage and Hour Administrator.
- Law / rule (federal, 29 CFR 785.48(b)): rounding to the nearest 5 minutes, tenth of an hour or quarter hour is permitted only where it averages out so employees are fully compensated over time. A closeout objective should include checking that an AI-drafted time summary has not rounded hours in one direction.
- Law / rule (federal, 40 CFR 171.201, binds certified applicators who rely on noncertified applicators to use restricted use pesticides under direct supervision, and employers of commercial applicators): commercial applicators must keep records of each noncertified applicator's qualifications for two years, including the trainee's name and signature, the date of training, the trainer's name and a description of the training. States may impose stricter rules. An AI-drafted training record that drops a field fails that rule however tidy it looks.
Best practice: write the identification objective with humility built in. A tool's name for a weed, pest or disease is a hypothesis. The behaviour is to record it as unconfirmed, photograph the evidence, and route anything that would lead to a treatment to the certified applicator. Professional guidance (voluntary): NIST AI 600-1 (July 2024) describes confabulation as "confidently stated but erroneous or false content"; a confident name is not a confirmed one.
Best practice: one objective per session. The table has five rows. A single session should usually carry one, perhaps two. The others become later sessions.
Remember this
- Write one objective per task family: estimating, scheduling, closeout, identification, reporting.
- Every objective names the record it changes and the person who owns the decision behind it.
- Time records and supervision training records have legally required contents; an AI draft that drops a field is a compliance problem.
- A tool's identification is a hypothesis; the learner records it as unconfirmed and routes any treatment question.
- No AI session objective asks anyone but the certified applicator to decide anything about a product, rate or re-entry time.
Landscaping example
The fictional company Rillwater Landscape & Irrigation wants a winter training block on AI for its office and crew leaders. The training lead, Saoirse Kettleby, starts by listing every AI-assisted task in the company and sorting them into the five families. She writes one objective per family, then chooses two for the first session: closeout and identification. Her closeout objective reads: before sending a drafted closeout note, the crew leader deletes any line the visit record does not support and confirms the hours in the summary match the time record without rounding in one direction. Her identification objective reads: when the tool names a weed, pest or disease from a photograph, the crew leader records the name as unconfirmed, keeps the photograph, and routes any treatment question to the certified applicator. For closeout practice she builds five fictional visit records for five fictional properties on Orchard Rise, each with an AI-drafted note and time summary. One summary rounds every crew member's day up to the next quarter hour for travel and down for the last job, which nets to less time than the records show. One note lists a hedge trim the crew did not do. For identification, she prints four photographs of fictional plants from the company's training pack with the names an AI tool gave them; two names are wrong. The company's fictional certified applicator, Barnaby Tresize, reviews the material and asks her to remove one sentence from the answer sheet that suggested which product would treat the weed. Saoirse removes it and replaces it with "route to the certified applicator". In March she checks thirty closeout notes. She finds two with lines the visit records do not support and no summaries rounded in one direction. She takes the two notes to their crew leaders privately and adds a third practice note to next year's pack based on the pattern.
Responsible-use warning
Never write "staff will use AI responsibly" as an objective; it covers every task and changes none. Never write an identification objective that treats a tool's name for a pest or disease as confirmed, or an objective that asks anyone but the certified applicator to decide on a product or rate. Never teach rounding as a way to tidy time records, and never suggest an AI-drafted training or time record is compliant because it looks complete: it must carry every required field.
Human-review requirement
The instructor owns the objectives. The certified pesticide applicator reviews any objective or practice item touching identification that could lead to a treatment, and any supervision training record. The person responsible for payroll or the company's employment adviser reviews any objective touching time records or rounding, and confirms how the company's own records meet the federal rules and any stricter state rule. The person who signs proposals confirms what an estimator may promise.
Practical exercise
Write and test five behaviour objectives for a fictional company. Use fictional records only.
- List at least eight AI-assisted tasks in a fictional landscape company and sort each into one of the five families: estimating, scheduling, closeout, identification or reporting.
- For each family, write one behaviour objective naming the role, the trigger, the action, the record it is checked against, and what happens on a mismatch.
- Beside each objective, name the person who owns the decision behind the task.
- For the closeout objective, list the time-record items that 29 CFR 516.2 requires for a non-exempt employee that your fictional summary must preserve, and note that state rules may add more.
- For the identification objective, confirm the learner's action is recording and routing, not choosing a treatment.
- Check each objective against the rule that no one but the certified applicator decides anything about a product, rate or re-entry time.
- Choose one objective for a first session and explain in two sentences why it comes first.
- Build one fictional practice item for that objective with at least two seeded errors, one of which is an omission.
- Name the reviewer for your practice material and what they must check before delivery.
Knowledge check
-
What are the five task families in this lesson?
Answer guidance: Estimating, scheduling, closeout, identification and reporting, each producing a record that someone relies on and that an objective can be written against.
-
What does 29 CFR 785.48(b) permit about rounding, and what must a closeout objective check?
Answer guidance: Rounding to 5 minutes, a tenth or a quarter hour only where it averages out so employees are fully paid over time; the objective checks an AI summary has not rounded one way.
-
What must a commercial applicator's record of a noncertified applicator's qualifications include under 40 CFR 171.201?
Answer guidance: The trainee's name and signature, the date of training, the trainer's name and a description of the training, kept for two years; states may impose stricter rules.
-
How should an identification objective treat a tool's name for a pest or disease?
Answer guidance: As an unconfirmed hypothesis: the learner records it as unconfirmed, keeps the evidence, and routes any treatment question to the certified applicator.
Sources and further reading
- AI Coalition Network — Certification Educational Standard: Human Review of AI-Assisted Professional Work. Accessed 2026-09-15.
- Wage and Hour Division, U.S. Department of Labor (via U.S. Government Publishing Office) — 29 CFR Part 516 - Records To Be Kept by Employers. Accessed 2026-09-15.
- Wage and Hour Division, U.S. Department of Labor (via U.S. Government Publishing Office) — 29 CFR Part 785 - Hours Worked. Accessed 2026-09-15.
- United States Environmental Protection Agency (Code of Federal Regulations) — 40 CFR § 171.201 — Requirements for direct supervision of noncertified applicators by certified applicators. Accessed 2026-09-15.
- United States Congress (Federal Insecticide, Fungicide, and Rodenticide Act) — 7 U.S.C. 136j — Unlawful acts. Accessed 2026-09-15.
- National Institute of Standards and Technology — Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1 (2024-07-26). Accessed 2026-09-15.