Imagine two messages reaching a tea shop on the same evening.
The first says: best decaf green tea.
The second says: I want a good decaf green tea for my mother. She likes lighter flavours, caffeine keeps her awake, she drinks two cups a day, she is on a tight budget and she would prefer organic if possible.
Both want decaf green tea. Only one has told me enough to help properly. The second tells me who the tea is for, when she drinks it and what she can afford. Two cups a day on a tight budget means price per cup matters more than the price on the box.
That second message is what I call Premium Intent.
What I mean by Premium Intent
Premium Intent, as I use the term, is a request that carries the whole situation: the context, the emotional stake, the timing, the constraints, the budget, what a good result would look like and what the person would rather avoid. Years of search boxes trained most of us to squeeze a need into three words. Talking to an AI assistant, by voice or by typing, lets the need come out at full length.
"Premium" describes how much the request tells you, not a fee and not a posh customer. I came up with the name in June 2026 while rebuilding teas.co.uk. Marketers talk about high intent and buyer intent, and the two words may well have met before. I don't know whether anyone has used the name this way, but it suits what we've built. My Tea Wiki entry notes when I started using it.
A person who explains their situation has already done much of the choosing, so an answer that respects what they said is more likely to fit and to bring them back. It also raises the bar. If someone tells you the truth about their situation, pushing whatever you would rather sell betrays that trust.
Three imagined days
When I explain the idea, I follow three of these requests from start to finish. I want to be plain about them. The days are real only in the sense that they are ordinary life. They are not my customers and none of it happened. They are three imagined requests, written to show the shape of the idea.
The father. Picture a man who has his son for two months this summer, after four years without two full weeks together. The boy loves Pokémon, and the father cares less about cost than whether the money makes the trip better for his son. A keyword search hears "Pokémon holiday". A good answer hears a father trying to close a gap. It spends more on a hotel near the big day, so the boy does not arrive exhausted, and less on a flight upgrade he would sleep through. As I put it: spend where it helps the boy, refuse where it does not.
The proposal. A man in London has the ring and a partner who cannot stand being the centre of attention. A good answer starts with her: a private box rather than the front row, a quiet table, no photographer, and a straight "Do not pay for the premium suite, you will be asleep most of the time you are in it." In that version of the day, tea has a small part, a calming tea waiting in the room for the morning after. That is the right size for a tea shop in someone else's big day.
The family. Two parents, an eleven year old and a six year old, plan New York, Las Vegas and the Grand Canyon by hire car without realising how far apart they are. A good answer will not let them plan a thirteen hour drive with a six year old in the back. It flies the long legs, drives the short ones, and raises what they did not know to ask, such as insurance and rest days.
Each answer is shaped by what the person said, and says where not to spend as clearly as where to spend more.
What is built, and what is not
None of those days can be booked through teas.co.uk. What I picture is hotels, theatres and restaurants each holding their own checked facts, with an assistant joining them into one plan. Most of that network does not exist yet. An assistant acting for a customer across several businesses belongs to version five of the build, which is planned, not built.
What is built is the tea end, and it is real.
The Ask API at teas.co.uk declares premium_intent_supported. It asks any assistant calling it to send the full human request, with the context, constraints, budget and what success looks like, not a keyword summary. The mother's decaf green tea should arrive as the whole sentence, not as "decaf green".
Behind that sits one checked set of product facts, live since version two launched on 10 May 2026. Every product has structured data behind its page, with identifiers, prices, availability, source links and price per cup. And since 6 October 2026, with the customer's approval, an assistant can search our catalogue, compare teas and fill a basket through the teas.co.uk MCP server. The customer checks the basket and pays on teas.co.uk.
I cannot yet tell you how many of the requests reaching us are long and rich ones, and I will not guess.
Why I opened the door
Much of what I read about search and AI is about blocking AI systems to guard content. I went the other way. teas.co.uk is open to every assistant that behaves well, and protected against the ones that do not, because I want as many AI systems as possible to know the shop. The more that know us, the more often we are cited, and citations bring visitors and, in time, orders.
Premium Intent is why that bet makes sense for a small specialist. A long, careful question is where depth beats size. A shop that has done the work on its facts can answer the tea part of a request better than a catalogue of everything.
Those three days have not happened. The request they describe is already the kind we ask assistants to send us.
What others have found
Outside research and reporting that bears on this, with links to the original sources.
- Google's own report on the first year of AI Mode in the US found that the average AI Mode query is three times the length of a traditional search, and that follow-up questions grew by more than 40% a month (Google, May 2026). Search Engine Journal covered the findings in Google data shows AI search users moved past keywords and AI Mode queries are 3X longer.
- In the same Google data, planning queries grew 80% faster than AI Mode queries overall, and comparison questions beginning with "which" grew 40% faster (Search Engine Journal).
- OpenAI launched Instant Checkout in ChatGPT in September 2025, built on an open standard called the Agentic Commerce Protocol (OpenAI). Search Engine Journal later reported that Instant Checkout gathered roughly thirty Shopify merchants in six months before it was retired (Search Engine Journal).
- Slobodan Manic argues in Search Engine Journal that agentic commerce is a structural change rather than another voice shopping fad, but that it is still unproven (Agentic commerce is not the next voice shopping, but it is still unproven).
Read more
- Premium Intent
- The complete payload
- Claim boundaries
- Built versus planned
- Wide versus deep
- Lee Samuel Tucker's builder vocabulary
- Meet the founder: why I built teas.co.uk
- Lee Samuel Tucker on the Tea Wiki
- Under the Hood
More from the news
- Why I call it Premium Intent
- The decaf fruit tea problem
- What's inside a tea shop built for AI to read
Lee Samuel Tucker
Founder, teas.co.uk
Tunbridge Wells, Kent
