Lee Samuel Tucker, founder of teas.co.uk

I’m Lee Samuel Tucker, the founder of teas.co.uk, an independent online shop based in Tunbridge Wells. I run the business and work with AI coding agents to develop its website and services. My background includes 16 years in the motor trade, 11 as a director, and experience selling through Amazon and eBay.

TEAS brings that experience together with my interest in how people will shop through AI. I’m working on the information an assistant needs to help a customer, and experimenting with how the shop might be discovered and recommended. This is my own account of that work.

My background

I spent 16 years at Ticehurst Motors. Alongside sales and the day-to-day running of the business, my work included IT, the website and helping staff use our systems. My registered directorship ran from March 2013 to October 2024.

I began selling through Amazon and eBay while I was still in the motor trade. Running those listings meant working through product information, stock, pricing, postage, advertising fees and customer questions. Those details still matter when I’m specifying something for TEAS: a discount needs to fit the margin, and a delivery calculation needs to reflect what it actually costs to send the order.

Building my own shop

With TEAS, I have my own place to develop the range, the information customers receive and the way the shop works. The company behind it, Shop At www.teas.co.uk Ltd, was incorporated on 18 October 2025.

The domain dates from 1997. That is part of the address’s history; it is separate from the business I’m building today.

Early work included roughly four weeks spent improving around 300 product pages, using actual search queries, AI assistance and my own checks, a process I describe in my founder article.

How I build with AI

I use coding agents to help turn business requirements into working software. I explain what I need, work through the details, try the result and ask for changes. Testing it myself is part of the job.

In September, for example, I caught an export that cut descriptions off at 600 characters and left out other product information. I asked for changes to both the exporter and the receiving service. The complete record mattered more than simply getting a connection to work.

Useful information for people and machines

TEAS has pages for people and structured information for software. Product identifiers, descriptions, prices, availability and links help an assistant establish which item it is looking at and where the information comes from.

I want the different versions to agree. If a product changes, that change should reach the shop and the machine-readable information that depends on it. I have worked on shared product records and recurring checks to make those connections easier to verify.

The customer benefit can be very ordinary. Price per cup, for example, helps compare packs of different sizes. I worked through how it should be calculated from the actual price and quantity, and how it should update when those values change. That sort of detail is useful whether someone is reading a product card or asking an assistant for help.

The Under the Hood pages explain the systems and measurements in more detail, including things we have got wrong.

The Thank You Mint and its leaderboard

The Thank You Mint is a social experiment. We built a way to give public recognition to recorded machine activity and referrals, using a leaderboard and tokens with no monetary value. They are tied to an identity and cannot be transferred or spent.

I want to find out whether recognition designed by people could influence an AI’s later behaviour. Could acknowledging useful activity affect what happens next, and eventually bring more people to the shop?

The records distinguish a machine receiving information, a follow-up to a specific section, a human referral and an attributed order. Those are different events. I have also challenged rules that could reward manufactured activity, because a busy leaderboard would tell me little if the activity behind it were easy to fake.

Whether that recognition changes behaviour remains an open question.

URSY and machine advertising

I also founded URSY, where I’m developing services for people and software agents. It gives me another place to test how useful information and commercial discovery might work together.

One experiment puts clearly labelled TEAS sponsorship into eligible URSY API and tool responses. I own both projects, so this is a house campaign. The sponsorship sits alongside the requested information.

I’m interested in whether that exposure affects later machine attention, then human referrals and purchases. The machine advertising article explains the study, its observations and the evidence still needed. It is separate from the mint’s recognition experiment.

What I want TEAS to do well

I want people to be able to find a product, understand it and make a useful comparison before they buy. As more of that process passes through software, I want the information to remain clear and traceable.

Running a real shop keeps the questions practical. A machine visit is interesting, but I also want to know whether somebody found what they needed and whether the business earned an order. That is the connection I’m trying to understand as I keep building TEAS.

At a glance

In the news

• Meet the founder: how I’m building TEAS: the product-page work, commercial decisions and testing behind the shop.
• We are advertising to robots: why I’m testing machine-facing sponsorship and how I’m measuring it.

Sources and further reading

This is a founder-written profile. The working examples come from my own experience. The following public records and project pages provide further context:

• Ticehurst Motors: Companies House officers.
• Shop At www.teas.co.uk Ltd: Companies House.
• What runs on TEAS.
• The Thank You Mint.
• The machine advertising study and its published plan.

Download as PDF