An AI agent for tradespeople isn't another piece of software, it's a role: a fixed task, an approval by you, and a limit that's set in advance. In a trades business, the desk work that comes back the same way every week fits that model: pre-sorting inquiries, drafting quotes, following up on open quotes, creating the invoice as a draft once a job is confirmed, matching receipts. What happens on site stays with people: the measurements, the prices, the conversation with the customer, and everything your business is liable for. Honest up front: I don't have a documented case from a trades business. This page shows which roles from my own workforce transfer to a trades business, and how.
I run my business with a workforce of fourteen AI employees. What an AI employee fundamentally is and how one comes together is collected on the page AI employees: tasks, costs, getting started. Two parts of the trades question I've already covered in depth, and both are the foundation for this page: Automating Quotes in the Trades, with your list of standard services, materials versus labor, and lump sum versus unit price, and Winning New Customers in the Trades, with response speed, referrals, map listings, and reviews.
Roles, not a tool list
Search for "AI in the trades" and you mostly find lists along the lines of "the 7 best AI tools for tradespeople". Lists like that answer one question: which software exists. They don't answer who in your business does what with it, who approves the result, and where the AI's work stops. That's exactly where adoption fails, not at picking the software.
So I go about it differently. For every task an AI is supposed to take over, three things are written down beforehand:
- The task: exactly what it does, with which inputs.
- The approval: what you check before anything leaves the business.
- The limit: what it never does, even if it technically could.
That's the difference between a tool and an employee. A tool can do a lot, an employee has a job.
What comes back every week in a trades office
The right-hand column isn't shorter because the AI can do so much, but because that's where the decisions sit. The left column is diligence work, the right column is your trade. How appointment proposals come together with AI in general is in Scheduling Meetings With AI.
The quote: where time and risk meet
Among these tasks, the quote deserves a closer look, because this is where time and risk meet. A complex quote can sometimes take two hours, and much of that is searching through old quotes and price lists. That's where the mistakes happen: a forgotten line item, an outdated material price, a time allowance that's too tight. If a quote like that goes out and the customer accepts, a job can quickly turn into a loss.
That's exactly where an AI employee can come in:
- From your history: it finds comparable earlier quotes, carries over the matching line items with units and can build a first draft from them within a few minutes. Time allowances and prices come from your service list, the quantities from your measurements.
- With current prices: it pulls prices from your current price source, not from the old quote, and flags what has changed noticeably. If a price is missing, it asks.
- With a checklist: missing standard items, mismatched units, totals, placeholders, each appears as a note next to the draft.
- Ready for sign-off: your office no longer gathers anything, it reviews and decides. Only what you have approved goes out.
That saves time, and above all it lowers the risk every quote carries when it leaves your business. The prerequisite remains a maintained service list, as described below. Once it's in place, you should definitely tackle the quote. What this looks like in detail, with the service list, price connection and checklist, is in Automating Quotes in the Trades.
Which roles from my workforce transfer
The job descriptions below are those of my own AI employees, as they appear on their pages. They work for my business, not for a trades business. What transfers is the pattern: task, approval, limit.
Anton, Sales. He drafts proposals as a PDF and, once a deal is confirmed, creates the contact, quote and invoice as a draft in the accounting tool. In the trades, that's the most obvious role, with one prerequisite: without a maintained service list with units, time estimates, and prices, there's no usable draft. How to build one is in Automating Quotes in the Trades. Anton's rule is fixed: no price source, no price, and no quote leaves the system without your click. More on Anton, my employee for quotes and invoices.
Frieda, Office inbox. She sorts incoming mail into four stacks and drafts replies; nothing goes out unchecked. In a trades office, that means customer inquiries, supplier emails, and follow-up questions no longer land in one pile, and for inquiries there's an acknowledgment ready that asks for the details you need for the callback. How she works is on Frieda, my employee for the inbox, and a method to do it yourself is in Sorting Your Inbox: The Pile System.
Sonja, Office and assistance. She prepares new-client emails and keeps the inbox in order; nothing is sent without approval. Transferred to the trades, that's the follow-up email after the quote, the one that otherwise gets lost in the day-to-day. When and how to follow up is in How to Follow Up on a Quote by Email.
Olaf, Lead scout. He finds local businesses with an outdated website and delivers a rated list with a briefing. That fits businesses selling to commercial customers better than a business that gets inquiries from homeowners. The pattern behind it: search by fixed criteria, justify every find, and never contact anyone on his own. How Olaf works and what that means for winning customers in the trades is in Winning New Customers in the Trades and on Olaf's page.
Ralf, Website optimizer. He turns an old site into a modern one-page website on his own, live as a demo on request. For a trades business, the website is mainly proof: references, services, service area, a phone number you can tap. What pays off in which order is in Modernizing a Website With AI, and how Ralf works on Ralf's page.
Peter, Bookkeeper. He screens receipts in the accounting interface, sets receipt types and matches card statements. Peter isn't available as a package yet. The pattern still holds: preparing and matching receipts yes, booking and tax judgment no, that stays with you and your tax advisor. What AI takes over in bookkeeping today is in AI Bookkeeping: What Actually Works Today, and how Peter does it for me on Peter's page.
Conny, Content editor in chief. She drafts reels, hooks and captions for my Instagram channel, checked against the house rules. For a trades business, that would be the draft for a post about a finished job. You take the photos, and whether a customer's property can be shown is something you clarify with the customer, not the AI.
Apart from Peter, all of these roles are available as ready-made packages in my community.
What doesn't go to the AI
In the trades the boundary is clearer than in most industries, because a large part of the work doesn't happen at a desk at all.
Measurements. Someone on site takes them. An AI can carry the quantities into the quote draft once they've been recorded properly, but it can't take them.
Setting prices. Prices come from your costing, not from a model that guesses market rates. If a price source is missing or contradictory, the process calls for a question, not an estimate.
The conversation with the customer on site. Whether the customer trusts you is decided at the appointment, not in the email before it. An AI can deliver the preparation, not the conversation.
Liability questions. Defects, warranty, change orders, everything your business stands behind is your decision. An AI can assemble the case and sort the documents. The decision and the reply to the customer stay with you. What applies legally to change orders and cost estimates in Germany is in Automating Quotes in the Trades.
On top of that, one rule applies to every role: nothing leaves the business without you having looked at it. No quote, no invoice, no email to a customer.
Where a trades business should start
With one role, not seven. Which one depends on where your time gets stuck:
- If inquiries sit unanswered, start with the inbox. Before that, note for one week when each inquiry came in and when it was answered. That list shows you whether the problem really lies there.
- If your evenings go into quotes, start with your service list, not with AI. Only once items, units, and prices are maintained does a quote employee pay off.
- If quotes fizzle out unanswered, follow-up is the first role.
What an AI employee costs, I work through in What an AI Employee Really Costs: depending on the tier, the subscription runs roughly 20 to 200 euros a month (as of August 2026), plus your onboarding time. Whether an employee like that replaces office staff or takes work off their plate, I answer in AI instead of employees?. What the same roles look like in other industries is shown in AI Employees for Consultants and Agencies and in the real case from a tax firm.
Common questions
What is an AI agent for tradespeople?
An AI employee with a fixed task in the office, such as pre-sorting inquiries or drafting quotes. It works from a written description, puts results down as drafts, and sends nothing without your approval. The difference from a chat tool is explained in What Is an AI Employee?.
Which AI tools do I need as a tradesperson?
Fewer than the lists suggest. More important than the tool are the foundations: a maintained service list, an accounting tool where invoices are created, and an inbox where inquiries arrive. An AI employee works on top of those foundations, it doesn't replace them.
What does an AI employee cost a trades business?
Depending on the tier, the subscription runs roughly 20 to 200 euros a month (as of August 2026). The bigger item is your time: writing down the process, maintaining the service list, giving feedback. The full calculation is in What an AI Employee Really Costs.
Is there an example from a real trades business?
I don't have a documented example with numbers from a trades business, and I'm not going to make one up. What I can show are the roles from my own business and a real case from a tax firm, where the path to sign-off works the same way: clean runs first, then watching in parallel, then running alone.
Does AI replace office staff in a trades business?
It takes diligence work off their plate: sorting, drafting, keeping track. Decisions, customer conversations, and anything involving liability stay with people. The general answer is in AI instead of employees?.
How to take this further
Write the six tasks from the table above on a sheet of paper and next to each, how many hours a week it takes at your business. Mark the one with the biggest share whose result you can check fastest. That's your first role. How an AI employee comes together from picking the task to sign-off is in Hiring an AI Employee: The Process. The ready-made packages, Anton and Frieda included, are in my community Claude Practitioners, and you can bring your service list or your inbox into a call there.
