Naming, Mood-Words & Language

Naming a project - a building, a housing scheme, an interior collection, a practice - is a task with a strange shape. Producing candidates is almost free, and Claude makes it freer: you can have thirty names in one reply, forty more in the next, each cluster in a different direction. But the value is not in the candidates; it is in the choice, and the choice is entirely yours, because a name has to carry meaning you understand, survive a client saying it out loud, and fit a story only you know. Claude can generate the field; it cannot pick the winner, because it does not know what the name has to do.

So treat Claude as the ideation engine and yourself as the editor, and lean into the asymmetry. Ask for far more than you need, in deliberately varied registers, so your final choice is made against real range:

I need names for a boutique homestay in a restored Chettinad mansion -
warm, rooted, not twee, works for Indian and international guests.
Give me 30 candidates in five groups: (1) Tamil words tied to the
house, (2) references to the courtyard/light, (3) the family/heritage
angle, (4) one evocative English word, (5) wildcards. One line each on
what it evokes.

Grouping matters more than volume. Thirty names in a flat list is noise; thirty names in five labelled directions shows you the choices behind the choice - do you want to signal heritage, atmosphere, or lineage? Often the real value is not a name at all but the realisation, seeing them sorted, of which direction you actually believe in. Then you narrow. And crucially, when a name survives your shortlist, you have one more job Claude cannot do for you, covered in the next sections: checking that it is actually yours to use.

A reliable naming process is a funnel, drawn in the figure, and each stage does a different job. At the top, Claude generates wide - thirty to fifty candidates across varied directions. Then you pass them through filters that matter, and this is where your judgement does the work. Four filters catch most of it: does it mean something (or at least evoke the right thing); is it memorable and sayable (can a client repeat it down a phone, does it survive an accent); is it available (not already a well-known project, product or trademark in your field); and does it fit the story (would it sit honestly beside the rationale you wrote in the last lesson). What survives all four is a shortlist of three to five, from which you - not Claude - choose one.

You can enlist Claude for parts of the filtering, carefully. It is good at pressure-testing memorability and connotation: "For each of these five, tell me how it could be mis-said, mis-spelled, or mis-heard, and any unfortunate associations you are aware of." That surfaces the name that becomes a rude word when slurred, or reads badly in a second language - genuinely useful, and cheap to ask.

But note the asterisk in the figure, because it is the whole risk of this lesson. Claude cannot reliably check availability. It does not have a live view of trademark registers, company names or domains; if you ask "is this name taken?" it will produce a confident, plausible answer that may be completely wrong - a classic hallucination, and a costly one if you name a scheme after a trademark someone already owns. Availability is verified by you, at the source: a trademark search, a company register, a domain lookup, a quick web search on your own plan if you have it. Use Claude to generate and to sanity-check feel; use the real registers to confirm you can actually use the name. The funnel narrows candidates; only external checks make a name safe.

Long before a name, and all through a project, you need language of a humbler kind: mood-words. These are the handful of adjectives and short phrases that fix the feeling of a scheme - unhurried, grounded, quietly formal, warm-but-not-soft - so that everyone touching the project, from your team to the client to the contractor picking a paint sheen, is aiming at the same target. Mood-words are how a design intention travels without a drawing, and they are especially powerful in interiors, where a scheme lives or dies on a coherence of feeling across dozens of small choices.

Claude is excellent at helping you find and sharpen this vocabulary, because it has read an enormous amount of writing about atmosphere and can offer precise words when your own reach for the obvious ones. Give it the feeling and ask for range and precision: "I want this apartment to feel calm but not cold, rich but restrained, Indian but not thematic. Give me twelve mood-words that capture that, and for each a one-line gloss so my client and I mean the same thing." The gloss matters - agreeing that "warm" means the amber of evening light, not the heat of a bright colour, prevents the slow drift where everyone nods at a word and pictures something different.

The figure shows a useful way to organise this for a scheme: a small vocabulary map across registers - material, light, mood, touch - each with a few chosen words. Held together, it becomes a brief in miniature: lime-washed, raw teak, brushed brass; dappled, low and warm; unhurried, grounded; cool stone, soft weave, worn edge. A client who reads that map feels the room before a single sample is bought, and your team can test any proposed finish against it - does this fitting read as unhurried and worn, or as slick and new? As always, you choose the final words: Claude widens your vocabulary and sharpens your definitions, but the feeling being named is yours, drawn from the scheme you designed.

The last layer is the short, load-bearing phrase: a tagline for a competition board, a one-line description for a website, the title of an interior collection, the sentence a developer will put on a hoarding. These are hard precisely because they are short - every word carries weight, and the difference between memorable and embarrassing is one syllable. Claude is a strong tagline partner because it can spin twenty variations on a single idea, each with a different rhythm and emphasis, letting you hear options you would not have written and then choose the one that rings true.

Brief it the way you would a copywriter: give the idea, the audience, the feeling, and the constraints, and ask for range within them. "Twenty taglines for a low-carbon office in Bengaluru, under seven words each, confident not preachy, no cliches like green or future. Vary the angle: some about the building, some about the people in it." Then apply the same funnel - meaning, memorability, fit, availability - and read the survivors aloud, because a tagline lives in the ear.

Two honesty checks close the lesson. First, meaning across languages: if you are drawn to a word in Tamil, Hindi, Italian or any language you do not speak fluently, do not trust Claude's gloss as final. It is often right and sometimes confidently wrong about nuance, register and connotation - the difference between a word that means "home" warmly and one that carries an unintended overtone. Check with a fluent speaker before it goes on a building. Second, cliche and sameness: because Claude gravitates to the statistical centre, its first taglines are often the ones a hundred other projects already use. Push it explicitly past the obvious - "those are cliches, give me twenty that avoid them entirely" - and do the final taste-check yourself. Generated language is a starting field, never a finished choice; the word that goes on the building is one you understood, checked, and picked.