Eliminating Watch Descriptions with AI Voice.

Tom runs The Wrist Watcher — a watch retail business out of London. Every watch he stocks needs a description before it goes live on the site: condition notes, specifications, history, character. The kind of detail that separates a serious watch dealer from a generic listing.

The problem was the process. Every time a new watch arrived, Tom would inspect it — then sit down and write the description from scratch. Fifteen minutes per watch, pulling details from memory, matching the tone of his existing listings, making sure it read like him. Multiplied across his inventory, it added up fast.

Time spent writing was time not spent sourcing, selling, or building the business.

We built Tom a voice agent that works the way he already works.

Now, when a watch arrives, Tom inspects it and talks. He describes what he sees — condition, references, details, anything worth noting — directly into a voice note on his phone via Telegram. The agent transcribes that note and turns it into a finished listing description, ready to go on the site.

To make sure it sounds like Tom, we trained the agent on every watch description he'd ever published. It learned his vocabulary, his cadence, the level of detail his customers expect. The output doesn't sound like AI copy — it sounds like Tom wrote it.

The finished description lands back with him for a quick review. He copies it across. Done.

Writing watch descriptions is no longer a task Tom does.

What used to take 15 minutes per watch now takes seconds of his attention. The hours that were locked up in typing are now spent on the parts of the business only Tom can do — sourcing watches, building relationships, creating content.

The best automation doesn't change how someone works. It removes the part of the work that shouldn't exist in the first place.

Tom is an expert in watches — sourcing them, understanding them, selling them. Writing the same kind of description, over and over, was never a good use of that expertise. It was friction dressed up as a task.

The right AI solution recognised the real workflow — inspect, describe, move on — and built around it. No new habits. No new tools to learn. Just minutes back, every single day.