Your customers shouldn't have to understand your menu to get what they came for.

A website that responds to intent lets the visitor say what they need — in their own words — instead of decoding your navigation to find it. Rather than clicking through menus and guessing which page holds the answer, the visitor states the intent ("I need to cancel and rebook," "something warm under fifty euros") and the site understands it and takes them to the exact section. Maya makes an ordinary site do this: it reads your site's structure so it knows where everything is, then responds to what the visitor means — by voice or text — and navigates them there for them.

Last updated 2026-07-16

Why a menu makes the customer do the translating

Every menu asks the visitor to translate their intent into your categories. They came to "change a booking"; your site offers "Account," "Support," "Policies" — and now they have to guess which one you filed it under. Most won't guess more than twice. The bigger your site, the worse the tax: the customer knows exactly what they want and still can't find where you put it, so they leave with the intent unmet and you never see it.

Maya removes the guessing. A menu-driven site is a vending machine — the customer has to know the slot number before it gives them anything. Maya is the shop assistant on the floor: the customer just says what they're after, and Maya — which has read the whole layout — takes them straight to it, scrolling and navigating for them. The customer no longer needs to understand your menu; they tell the website what they need, and it responds to the intent, not the click.

What it means for a site to respond to intent

The visitor states intent — Maya takes them to the exact section — "where do I cancel," "show me the vegetarian options," "I need the warranty terms" — Maya knows which part of the site holds each answer and navigates there for them, hands-free, no clicking through menus. The visitor gets where they meant to go without ever learning how you organised it.

It understands your site's structure — not just keywords on a page — Maya maps where things live — which page has the sizes, where the returns policy sits, which section lists the family rooms. So it responds to what the visitor means, not just words that happen to match, and takes them to the right place instead of the nearest keyword.

It refines the request conversationally when the intent is fuzzy — the visitor rarely knows the exact product name. "A gift for someone who cooks" → "under eighty euros" → "nothing they'd already own" — Maya holds each constraint and narrows toward the right option, then explains why it fits, from your live catalogue. It responds to the messy real intent, not a perfectly-typed search query. (Selling and refinement on the Booking plan and up.)

It works by voice or text — and drives the page either way — the visitor can talk to it or type to it, and Maya acts the same: it does the scrolling and navigating itself, hands-free — no mouse, keyboard, or screen — right up until they choose to book or pay. Even someone one-handed on a phone gets to the exact section by just saying what they want.

It responds to intent in the visitor's own language — and follows a mid-conversation switch — Maya replies in the language of every message, so the visitor states their intent in whatever language they think in — and if they slip from English into French mid-sentence, it follows without missing a beat. When your site already has that language, it switches to the correct version. Intent shouldn't get lost in translation, and here it doesn't.

The intent that would have left becomes a captured lead — the visitor whose need your menu couldn't surface used to just bounce, unrecorded. Now Maya captures name, need, and contact in the same breath — so an unmet intent turns into a recoverable lead instead of an anonymous exit.

It tells you the intents your site couldn't answer — Maya surfaces what visitors kept asking for that your pages don't cover — "14 people looked for airport transfers." You stop guessing what's missing; the gaps in what your menu can serve show up as a weekly list, from real intent.

What happens when Maya doesn't know the answer?

Responding to intent raises an obvious worry: what if it guesses wrong and just makes something up? Maya checks its answers against your approved business information before it responds. When it can't verify something — an intent your site genuinely doesn't cover, a detail you never published — it says so instead of guessing, offers the visitor the right next step, and flags the question for you. You can mark it correctly handled, or teach Maya the answer in one sentence.

What that means for you: Responding to intent is only trustworthy if Maya knows the edge of what it knows. It gets more useful over time without ever being rewarded for confidently inventing things — so when a visitor's intent runs past your real information, they get an honest answer and the right next step, never a confident guess that sends them the wrong way.