Key takeaways
- 58% of small businesses used generative AI in 2025, and 84% planned to increase their use of technology platforms[1].
- The best office uses are drafting and summarizing: emails, letters, job ads, report summaries and policy first drafts.
- A person checks every output before it reaches a customer, employee or the books.
- Set data rules first: no customer personal details, card numbers, passwords or employee records in public AI tools.
The office is where AI already lives
Most talk about AI in plumbing focuses on answering calls. Yet in many shops the first real use is quieter: an office manager asking a chatbot to tidy up an awkward email to a customer, or to turn a messy page of notes into a clear policy. That use is spreading fast. In a U.S. Chamber of Commerce survey, 58% of small businesses said they used generative AI in 2025, up from 40% in 2024 and 23% in 2023[1]. In construction, 19.8% of businesses said employees used generative AI for work tasks in a recent two-week period[2].
The gap between those figures is informative. Many plumbing office staff are using AI tools on their own, on personal accounts, without any company rules. That is not a reason to ban it. It is a reason to point the use at the right tasks and put a few guardrails around customer data.
Ten office tasks where AI helps
| Task | How AI helps | What a person checks |
|---|---|---|
| Customer emails | Drafts a clear, polite reply from bullet points | Facts, tone, any promise made |
| Complaint responses | Suggests a calm structure and wording | Every fact and the offer made |
| Estimate cover letters | Explains the scope in plain language | Scope, price and terms match the estimate |
| Job ads | Writes a first draft from your requirements | Pay, benefits and legal wording |
| Policies and checklists | Turns notes into a structured document | Accuracy against how you really work |
| Meeting notes | Summarizes a typed or recorded meeting | Decisions and owners are right |
| Report summaries | Explains changes in an exported report | Numbers against the source report |
| Bookkeeping prep | Suggests categories for unclear transactions | Bookkeeper approves every entry |
| Spreadsheet formulas | Writes or explains a formula | Test on known numbers first |
| Training material | Drafts quizzes and how-to steps | A senior person reviews |
Notice what is missing: anything where AI makes the final decision. Prices, credit decisions, hiring choices, payroll and tax positions stay with people. AI drafts; a person decides and sends.
Writing: where most of the time goes
Office staff in a plumbing company write constantly: replies to customers, notes to property managers, reminders about overdue invoices, explanations of why a job took longer than quoted. AI is good at turning rough notes into clear, friendly text. The trick is to give it the facts and ask for tone, rather than asking it to supply the facts.
- Good prompt: Rewrite this as a polite email to a homeowner. Facts: the replacement part arrives Thursday; we'll call Wednesday to confirm a time; no extra charge for the second visit.
- Weak prompt: Write an email explaining why the part is late. The tool will invent a reason.
- Good prompt: Make this past-due reminder firmer but still friendly. Keep the amount, invoice number and due date exactly as written.
- Good prompt: Shorten this to five sentences a busy property manager will read.
Keep a shared document of prompts that worked, so everyone in the office benefits from what one person figured out.
Bookkeeping and reports
Many accounting tools now suggest categories for bank transactions and summarize monthly results. These suggestions save time on routine items like fuel and supply-house purchases, but they are guesses. Have the bookkeeper approve every category, especially for anything touching payroll, sales tax, equipment purchases or owner draws. AI is not a substitute for your accountant's advice on tax treatment.
AI is also useful for explaining reports. Export a report from your field-service or accounting software, remove customer names, and ask for a plain-language explanation of what changed from last month. Then check every number it mentions against the original report. Our guide to getting reliable reports out of your plumbing software covers which reports to trust.
Hiring and HR, with care
AI can draft job ads, interview question lists and onboarding checklists quickly. Two cautions apply. First, check every pay, benefit and legal statement, because AI tools confidently produce wording that may not match your policy or your state's rules. Second, never paste employee records, medical information, performance notes or Social Security numbers into a public AI tool. If you use AI to help screen applications, a person should make every decision and you should be able to explain it.
The data rules to set first
The main risk of office AI is not bad writing; it is data going where it should not. Public AI tools may keep what you type, and personal accounts are outside your control when someone leaves. Larger businesses also carry legal duties: the California Consumer Privacy Act applies to for-profit businesses in California with gross annual revenue over $25 million, or that buy, sell or share the personal information of 100,000 or more California residents or households, among other tests[3]. Smaller shops should follow the same habits anyway.
- Never paste: card numbers, bank details, Social Security numbers, passwords, gate or lockbox codes, or employee records.
- Strip names and addresses from anything you paste; use Customer A or Job 1 instead.
- Use business accounts for AI tools, so access ends when someone leaves.
- Check each tool's settings for whether it trains on your inputs, and switch that off where you can.
- Keep AI tools out of anything involving customer payment processing.
AI also helps the people trying to fool you
The same tools that polish your emails polish criminals' emails. Phishing and spoofing drew 193,407 complaints to the FBI's Internet Crime Complaint Center in 2024, more than any other crime type it tracks[4]. Phishing emails used to be easy to spot by their spelling; they no longer are. Tell office staff that a well-written email asking for a password, a payment change or a gift card is no more trustworthy than a badly written one. Verify by phone, using a number you already have. Our article on fake invoice and payment fraud has a full procedure.
A one-page office AI policy
Approved tools
Name the tools staff may use for work, and require business accounts.
Never-paste list
List the data that never goes into an AI tool, and give examples.
Human review
Everything AI drafts is read and approved by the person who sends it, and that person owns it.
No final decisions
AI does not set prices, approve credit, make hiring choices or decide tax treatment.
Share and report
Share prompts that work; report mistakes without blame so the policy improves.
Is it actually saving time?
Office AI is easy to adopt and easy to overrate. A draft that takes ten seconds to produce and ten minutes to fix has saved nothing. Before deciding a tool is worth keeping, or worth paying for, run a simple two-week check:
- Pick three tasks the office does often, such as customer emails, past-due reminders and job ads.
- For one week, note roughly how long each takes without AI.
- For the next week, note how long each takes with AI, including the time spent checking and correcting.
- Count corrections that mattered: wrong facts, wrong tone, anything that would have embarrassed you if sent.
- Keep the tool for tasks where time fell and serious corrections were rare; drop it for the rest.
Most offices find the gains are real but uneven. Rewriting and shortening are reliable wins. Anything that requires facts the tool does not have, such as a customer's history or your actual warranty terms, tends to need so much correction that writing it yourself is faster. Knowing which is which is worth more than any single tool.
Example: an office manager's week
Example: on Monday the office manager pastes the anonymized week-ahead schedule into an AI tool and asks for a short list of days that look overbooked; she checks it against the board and moves two jobs. On Tuesday she turns the owner's voice-memo notes into a draft updated warranty policy, which the owner edits. On Wednesday she uses AI to soften a reply to an upset customer, keeping every fact the same. On Thursday she drafts a job ad for a second CSR and checks the pay range with the owner. On Friday she asks for an explanation of why a report shows fewer completed jobs than last month, then confirms the answer in the software itself. None of it involved pasting a customer's name or address.
Frequently asked questions
Is it safe for office staff to use free AI chatbots?
It can be for drafting and summarizing, if they follow clear rules: no customer personal details, payment data, passwords or employee records, and a person checks every output.
Can AI do our bookkeeping?
It can suggest categories and explain reports, but a bookkeeper should approve every entry, and tax questions belong with your accountant.
Should we tell customers an email was drafted with AI?
There is generally no need for routine emails, as long as a person has checked and sent it and every fact is right. Be honest if a customer asks.
What is the first office task to try?
Rewriting customer emails from bullet points. It happens daily, saves real time, and a person can check the result in seconds.
References
U.S. Chamber of Commerce Technology Engagement Center (2025). Empowering Small Business: The Impact of Technology on U.S. Small Business (2025) (opens in a new tab) โ
U.S. Census Bureau (2026). Business Trends and Outlook Survey (BTOS), response estimates by sector (opens in a new tab) โ
State of California Department of Justice, Office of the Attorney General (2026). California Consumer Privacy Act (CCPA) (opens in a new tab) โ
FBI Internet Crime Complaint Center (IC3) (2024). Federal Bureau of Investigation Internet Crime Report 2024 (opens in a new tab) โ