What AI can and can't do in a flower shop office

An AI writing tool is good at turning rough notes into tidy words. It's poor at knowing facts about your shop. That one split decides most of what follows.

A florist's office work is mostly short written jobs: replying to enquiries, describing bouquets, confirming deliveries, listing stock for the wholesaler. Wikipedia's floristry page splits the trade in two: retail florists serve consumers, and wholesale florists sell in bulk to the trade¹. Your office sits between the two. Plenty of those jobs are wording problems, and wording is what these tools do well.

Jobs where a draft helps

Think of the tool as a quick first-draft writer who has never seen your shop. It works well when you supply the facts and ask only for the wording.

• Turning a few bullet points into a polite reply. See replying to enquiries.
• Rewriting a stiff description so it reads like a person wrote it. See bouquet and product descriptions.
• Making a delivery or pickup message clear and short. See delivery and pickup messages.
• Tidying a messy list of sales notes into a table. See stock and supplier lists.
• Laying out a plan for a busy week. See planning for the big flower days.

In each case the facts come from you: the date, the price, the address, which stems are in the cooler. The tool shapes them into sentences.

Jobs better left alone

Some tasks carry more risk than they save in time.

• Anything that needs a fact the tool can't know. What's in the cooler today, what your wholesaler has available, and whether a driver can reach a street by noon are things only you know. A tool will still write an answer that sounds sure.
• Claims about the flowers themselves. How long a bouquet will last, or whether a stem suits a certain home, should come from your supplier or your own records.
• Private customer details. The Australian privacy regulator would rather you kept customers' personal information out of public AI chat tools². Names, phone numbers and delivery addresses are personal information. Read the OAIC's page for your own situation, and if you're elsewhere, check your own country's privacy regulator.
• Hard conversations. A complaint about a late wedding delivery, or a sympathy order, needs a human voice and often a phone call.

Why every draft needs a check

The maker of one well-known chat tool says plainly that it can produce incorrect or misleading outputs³. Wrong dates are on its list of examples³. For a florist, a wrong date or street is the whole job gone wrong.

The same page says to treat the result as a first draft and not a final source³. That's a good rule for the whole shop. Checking drafts before you send them lists the slips to look for.

Questions to ask before you pay for anything

Since this site takes a buyer's view, here's what to look at before you spend money on a tool or plan.

1. What exact job will it do each week? Name it. The OAIC cautions against adopting a product just because it exists².
2. Who will read each draft before it goes out? The OAIC also asks businesses to plan where a human fits into the work². Here, that's usually one named person.
3. Where does the customer information go? Read the maker's privacy page and terms before pasting anything personal.
4. What does the plan cost, and when was that price checked? Prices change often. Write the date next to the figure and cite the maker's own page, not a review site.
5. Can you stop? Pick something you can cancel without losing your shop's records.

Start with one job, such as enquiry replies. Run it for a few weeks with every draft read before sending. Then decide whether to add another.