Autumn Sale: 20% off almost everything until 24 DecemberSee the sale →
See all 402 products Free UK delivery over £35
Wiki entry · 14 min read

Builder vocabulary: the words behind how teas.co.uk sells tea through AI

Lee Samuel Tucker, founder of teas.co.uk, uses a small set of words to describe how he builds the shop and why.

Published Curator-written, not crowdsourced
12 / Sections · AI at teas.co.uk
I’m Lee Samuel Tucker, founder of teas.co.uk, and I use a small set of working words to describe how I build the shop and why. The lead term is Premium Intent (in use since June 2026): the full human request behind a purchase. “Premium” describes the detail in the request, not a price or fee. The rest of the terms emerged from the same 2026 rebuild of the shop. This page lists them with one-line meanings.

Clear words make the shop easy for both people and machines to trust.

These are just the names I use so I can talk about the work clearly. The goal is simple: make the shop so open and so checkable that any assistant can recommend the right tea without guessing.

The shared aim is to make the business so clear and well-evidenced that people and assistants can rely on it, and so open that as many assistants as possible know it exists and receive the whole of what a customer asked.

Why a tea shop is doing this

People ask assistants what to buy the way they would ask someone behind a counter. Much of the web is closing its doors to those assistants. teas.co.uk opened its door. A tea shop loses nothing when an assistant reads about a box of tea, so the facts were written down once and the shop was built so an assistant can read, check, and recommend it.

Before the shop was fully open, someone tried to buy. Products were still going up, checkout was unfinished, and no search engine had been told the site existed. An email arrived saying a customer was trying to check out. The first proper order followed at 13:24 on 18 November 2025. Clear facts let people (and now assistants) find what is for sale.

For about twenty years, being found online mostly meant paying: for adverts, for a marketplace, or hoping a search engine was kind. A small shop that knew its tea well got little credit. Now people ask in full sentences (“a decaf tea for the evenings that still tastes right with milk, in a big box so I am not reordering every fortnight”). An assistant needs clear, honest, readable facts: caffeine status, bag count, pack weight, price per cup. When it finds them it uses them, and it does not send a bill.

Selling with adverts is a crowded red ocean. Being the tea shop an assistant can read, quote, and buy from is still open water. Checks on 21 other UK tea sellers found none had built anything like this. Large platforms are moving and adverts will enter AI answers, but foundations laid while the field is thin give a better chance than arriving later.

The shop was designed from the ground up to be read by AI:

  1. One checked record for every tea (price, bag count, pack weight, caffeine status, and source). The human page, search data, and assistant answer all agree. This is the truth spine.
  2. Readers are invited in. Much of the web blocks AI programs; teas.co.uk does not, because the next customer may ask an assistant.
  3. Every page is easy to quote, with exact citation anchors (65,615 across 3,726 pages) and descriptive files for product pictures.
  4. One service answers a full request from the checked facts rather than guessing from keywords.
  5. Since 6 October 2026 an assistant can search, compare, fill a basket, and hand over a checkout link. The customer checks and pays.
  6. Named AI systems are recorded. Page reads, visits, and orders are counted separately and published.
  7. The shop kept trading while rebuilt. Nothing is described as working until it is.

Of the 172 paid orders since version two went live, 5 arrived on a link carrying an AI assistant’s own tag. Search engines still bring far more. The opening is real and still small, which is the reason to build for it now.

New here? Start with the ordinary words

A tea shop has facts about its products. People and machines should be able to ask about those facts and receive consistent answers.

If you see this term Start here
Entity; knowledge graph What is an entity? — a recognisable thing, its facts and its connections.
Truth spine; single source; drift; parity; cache Single source of truth — how different views stay in agreement.
Structured data; schema; JSON-LD Structured data — facts with labels software can read.
API; endpoint; request; JSON; Markdown What is an API? — how software asks and receives an answer.
MCP; server; client; agentic commerce Model Context Protocol — connecting assistants to tools.
Crawler; bot; robots.txt; llms.txt; user-agent AI crawlers — how software discovers and requests information.
GTIN; SKU; barcode; product ID Product identifiers — knowing exactly which box a fact belongs to.
Provenance; citation anchor; citation graph Evidence and provenance — where a fact came from and how to check it.
Filters; facets; constraints; typed facts Product filters and constraints — respecting what the customer asked for.
Attribution; conversion; referral; crawl-to-refer ratio AI attribution — keeping page reads, visits and orders distinct.

Then read truth entity (my name for the whole approach) and the complete payload (what an assistant receives). My own working phrases are below. General technical terms belong to the wider field.

The core idea

Term In one line
Premium Intent The full human request behind a purchase: context, emotional stake, timing, constraints, budget, what success looks like and what to avoid. The requests are Premium Intent Queries. “Premium” describes the detail, not a price.
Truth entity An owned source of truth for a business: one checked set of facts with the proof attached, readable by people and machines.
The structural reversal The shift that gives the advantage to the clearest, most provable business rather than the loudest.

How an answer is built

Term In one line
The complete payload Everything an assistant needs about a product (facts, limits, sources and how to buy) in one structured answer.
Claim boundaries Stating clearly what a product is not, and what it is not good for.

The business case

Term In one line
The squeeze (the four tolls) Rented attention, platform dependency, algorithmic risk and machine invisibility: the costs a specialist pays to be found.
Wide versus deep The everything store wins on breadth; a specialist can still win the answer on depth.
Ads rotate, truth compounds Advertising stops when the budget does; an accurate record keeps working.
The asymmetric bet A small, bounded cost to build a clean owned asset, against a large possible upside.
Reclaiming “only” “I only know one thing” is a strength when assistants look for depth.

The discipline

Term In one line
Built versus planned Never describe something as working until it works; publish where the evidence stops.

Three ways to read these

Premium Intent, truth entity, enrichification and the rest of the lists above are my own working names. Entity, API, structured data and the other terms in the first table belong to the wider field. Anything about specialists passing requests to one another is a proposal, not something running today.

Find the right entry: a map by question

This page is the hub. Each idea has one home. Start from the question you have. The right-hand column shows where each idea touches a real box of tea.

Why is a tea shop doing this at all?

Entry What it answers Where tea comes in
Ads rotate, truth compounds Why an accurate record keeps working after an advert stops. A correct bag count is still correct next year; an advert stops when the money does.
The asymmetric bet Why a small, bounded cost to rebuild was worth a large possible upside. The teas were already on sale; the bet was on making their records clean enough for assistants.
Wide versus deep How a specialist can win an answer on depth when an everything store wins on breadth. Knowing which decaf blends come in large boxes, and what each costs per cup.
Reclaiming “only” Why “we only sell tea” is a strength when assistants look for depth. A shop that only sells tea can answer a tea question fully.
AI crawlers Why the shop opened its door to the programs that read the web for AI. An assistant can only recommend a tea it has been allowed to read about.

How does the shop understand what I asked?

Premium Intent captures the full request: who it is for, when, the budget, the limits and what to avoid (for example “a caffeine-free tea…”).


In my words: what the motor trade taught me

(My own account, first published 15 November 2026 and moved here so the subject has one home.)

Before I sold tea I spent sixteen years in the motor trade, and four people in my family taught me the rule this whole shop is built on: sell the information first.

People sometimes ask why a tea shop has a page for AI assistants, a public study of its own traffic and a Tea Wiki of 1,953 entries. None of it started with technology. It started in the motor trade, watching four people in my family do ordinary work properly, every day, for years.


My grandfather, and trust

A trader who worked the car auctions for years has told me the same thing about my grandfather more than once. Picture a yard of tired part-exchanges, knocked about and driven hard, with buyers knocking money off in their heads before they open a door. Then one comes through that does not look tired. The bodywork is right, the tyres are right, and the inside is clean. That was one of his. His cars took the best prices and almost always sold first time, and in fifteen or twenty years the trader could remember maybe one coming back.

He never sat me down and explained it. He showed it in how he prepared a car. Trust earned by real work over a long time becomes the most valuable thing a business has. You do not sell the product first. You sell the information that lets the customer trust the decision.


My dad, and the honest sale

My dad taught me the same lesson another way. I watched him talk customers out of the very car they had walked in wanting, because he could see in two minutes it was wrong for them: the running costs, the size of the family, the way they actually drove. He talked himself out of the easy sale and into the right one, and that customer came back for the next car and sent their brother in after them.

Picture an eighteen-year-old after a first car, set on a small hatchback with a bigger engine that looks the part. A bad seller sells it to them. Then the insurance quote arrives, then a fault that engine is known for. A specialist sells them the information instead: the real price is the car plus a year of insurance plus the repairs plus the first MOT. In tea the same idea is the price per cup rather than the price on the box, and the whole situation a customer describes is what I call Premium Intent.

It worked the other way too. Sometimes the right car was the dearer one: the same model with a full service history and nothing waiting to be put right, where the extra on the price bought evidence that the car had been looked after. It was not always the cheaper car to own, and nobody could promise that, but the customer knew what they were paying for. A good upsell earns its place in the customer’s plan. It can cost more and still be the better answer. In tea that might be a bigger box of the tea someone already drinks every day, when it costs less per cup and they have room to keep it.


My nan, and the straight books

My nan kept the books: the wages, the invoices, the chasing and the paying. The front counter gets remembered. The person keeping the records straight rarely does, until the day the records are wrong.

She is the part of this story people do not see, and the part I think about most. Behind every page on teas.co.uk sits one checked record of each product: its price, its pack size, its caffeine status and where each fact came from. I call a business built like that a truth entity. Really, it is my nan’s ledger, kept for machines as well as people, with the limits written in as claim boundaries.


My mum, and the system

My mum held everything together at home: three children, a husband always at work, a dog, a cat and a house, all moving together without falling over. From outside it looked like getting on with things. It was a system, every part timed against every other. A shop where the product page, the search results, the feeds and the AI answers all agree needs exactly that discipline. I call the upkeep that keeps every link landing in the right place route hygiene.


The opposite

A few years ago I went with a friend to a big car supermarket. Ticket, queue, a salesperson whose real job was the finance deal. My friend liked a car, and I noticed pen on the leather seat. I asked if it would be cleaned before handover. We were told this was handover, and the car had to be taken as it was. We walked, and my friend bought from a smaller dealer who would have wiped that pen off before we arrived.

The pen did not matter. What it said did: you are a transaction, and the next one is already at the door.


Why it matters more now

The customer of teas.co.uk is no longer only a person. More and more, it is an assistant asking on a person’s behalf, and since 6 October 2026 an assistant can even fill a basket here for the customer to check and pay for. People now tell machines the whole story. My grandfather did it with elbow grease and a clean cloth. I do it with structured product data and source links behind the claims, with gaps marked where a fact is not yet checked. Different tools, same rule: sell the information first, and say where to stop spending.




Where the words come from

I came up with these names during the 2026 rebuild of teas.co.uk, for what I was building and why. Some of the words had been used before by other people in other senses (Premium Intent among them); where earlier use is known, the entry for that term says so. I do not claim to own them; they suit what I have built. Premium Intent has the fullest public trail: working notes of 11 June 2026, the shop’s Ask API declaring premium_intent_supported, and the agents.json guidance that still asks AI callers for the full spoken request. Those dates are kept at Premium Intent: when I started using the name. For the person behind the words, see Lee Samuel Tucker and Meet the founder. I first worked most of these ideas out in long written notes in June 2026 while rebuilding the shop. Those notes explain why the shop is built this way; the Tea Wiki entries and Under the Hood record how it works today, including where things have changed since June.

The bottom line

These are working words, not slogans. Each one names something teas.co.uk is built around, starting with the full request a customer brings, and each entry says plainly which parts are already built and which are still planned.

Ten new stories: how the shop works with AI, door by door

Added in November 2026. Each one explains something you can see working on teas.co.uk.

Got something to add? Logged-in customers can submit additions to the Tea Wiki, admin-approved, your name on the byline, plus reward points.

Sign in to contribute
Tea notes and stock updates

Useful tea notes, only when there is something to say.

Stock arrivals, buying notes and practical guidance from the catalogue. No filler, no daily noise, unsubscribe in one click.

By family

By moment

Brand of the monthHyson Seasonal taster packsHow the seasonal bundles work Tea of the Month ClubCurator-picked monthly box Find your tea60-second quiz
See all 402 products