I’m Lee Samuel Tucker. I run teas.co.uk, and I use AI to help me build it. My background is in running a business, looking after customers and selling products. That experience shapes the questions I ask when I’m working on the website: does this help someone choose, is the information complete, and does it actually work?
My entry in the Tea Wiki gives the background. Here, I want to explain some of the work behind the shop.
From cars to tea
I spent 16 years at Ticehurst Motors, including 11 as a director. My registered directorship ran from March 2013 to October 2024. Alongside sales and the day-to-day business, I looked after IT, supported the website and helped staff use the systems we relied on.
I started selling through Amazon and eBay while I was still working in the motor trade. The work overlapped. I was taking what I knew about running a business into online retail, with its own practical problems to solve.
Postage, marketplace fees, advertising and discounts all come out of the same sale. When I work on pricing tools, those details have to be included. A listing can attract orders and still lose money if the calculation is wrong.
The detail behind a product page
teas.co.uk became my own place to work on those problems. The company behind the shop was incorporated in October 2025. The web address is much older; the business I’m building on it is young.
One early job took me about four weeks: working through roughly 300 product pages. I brought together the existing descriptions and the questions people were using to find the site through Google. I used AI to help rewrite the content, checked its suggestions and updated the pages myself.
That work made the connection between search and shopping very practical. People arrive with a question. The page needs enough useful information to help them decide what to buy.
Price per cup is another example. I worked through how to calculate it from the price and the quantity in a pack, so the comparison could update with the product. Two boxes can have very different prices and still represent similar value. Showing the price per cup helps make that clearer.
Building with AI means checking the result
AI is part of my development process. I describe what needs to happen, work through the details with coding agents, test the result and ask for changes when something fails.
Sometimes the problem is obvious once you try it. While testing a shopping-app prototype, I found that the product cards looked ready but could not be opened. A search result needs to take someone through to the information they came for.
Other problems sit between systems. In September, I spotted that a teas.co.uk product export was cutting descriptions off at 600 characters and leaving out attributes and extra product tabs. I asked for the full record to be carried through, with changes to both the exporter and the receiving service.
Those are the sorts of checks I spend time on. A convincing screen or a successful connection is only part of the job.
Making teas.co.uk useful to an assistant
I’m also building teas.co.uk for software that helps people shop. Alongside the pages people read, the site provides structured product information, source links and ways for an assistant to retrieve the relevant facts.
I want a change to a product record to reach the places that use it. That means checking the connections between the shop, its machine-readable information and the related content. I have asked for recurring checks because those connections need looking after as the site changes.
The Under the Hood pages explain what runs and what the measurements can show.
Two experiments I want to understand
The Thank You Mint is a social experiment: could public recognition influence an AI’s later behaviour? We built tokens with no monetary value and a leaderboard to acknowledge recorded activity. I want to understand whether that might affect attention and referrals.
Through URSY, which I also founded, I’m testing labelled teas.co.uk sponsorship inside useful machine-facing responses. Could that eventually help people find the shop? The machine advertising article explains what we are testing and measuring. Both questions remain open.
A machine reading a page, a person following a recommendation and a customer buying something are different events. I want to understand the whole path, including where the evidence stops.
That is the work behind teas.co.uk: running the shop, improving its information and testing ideas about how people may find products through AI. The practical question stays with me throughout: what does this actually do for the customer and the business?
More from the news
- Biscuits are back: teas.co.uk is one shop for tea, coffee and biscuits
- 5,000 visitors from Google in 28 days, and our best month yet with five days to go
- teas.co.uk is older than google.com, by 173 days
- We are advertising to robots
Lee Samuel Tucker
Curator, teas.co.uk
Tunbridge Wells, Kent
