From your URL to a working bot, in plain English.
No jargon, no black box. Here's what actually happens.
1. We read your website
You send us your URL. Our crawler walks your public pages the way a search engine
would — following your sitemap if you have one, otherwise following links from your
homepage. It respects robots.txt and stays on your domain.
You can add anything else that lives outside your site:
| Format | Examples |
|---|---|
| Web pages | services, about, FAQ, contact, policies |
| PDFs | price lists, menus, brochures, terms |
| Word documents | policies, service descriptions |
| Text & Markdown | anything you've written up internally |
Then we show you the list of everything we ingested. You confirm it's all material you're happy for customers to see. Nothing goes live before that.
2. We turn it into answers, not a search index
Your pages get broken into passages and indexed by meaning, not just keywords. So a visitor asking "can I bring my dog?" finds the paragraph about pets even though it never uses the word "dog".
When someone asks a question, we find the handful of passages most relevant to it and hand only those to the AI, along with strict instructions to answer from them and cite them. That's the whole trick: the bot isn't recalling something it learned months ago, it's reading your current content and summarising it.
3. We tune it for your business
- Brand — your colours, your greeting, your bot's name.
- Tone — formal for a law firm, warm for a cafe. Your call.
- Boundaries — anything it should always decline or redirect.
- The handoff — where it sends people when it can't help: an email address, a phone number, or your contact form.
Then we test it against real questions from your business, and against a suite of known manipulation attempts, before a customer ever sees it.
4. You paste one line
<script async src="https://skrychats.com/widget.js" data-key="pk_live_your_key"></script>
Put it before the closing </body> tag. That's the entire
installation.
Where it works: WordPress, Squarespace, Wix, Shopify, Webflow, Ghost, and any custom site. If you can add a tracking pixel, you can add this.
What it does on your page: adds a chat button in the corner. Tapping it opens the chat in an isolated frame, so it can't affect your site's layout, styling or code — and your site can't break it.
What it doesn't do: no plugins to install, no account for your visitors, no advertising cookies, nothing for you to maintain.
5. It keeps itself current
We re-read your website every week. Change your opening hours on your site and the bot follows — you don't have to tell us, and there's nothing to re-upload.
If a page hasn't changed, we skip it, so the upkeep costs you nothing extra. Changed something important and don't want to wait? Ask and we'll re-read it that day.
6. You find out what people asked
Every month you get a summary: how many conversations, the most common questions, and — most usefully — the questions the bot couldn't answer.
That last list is usually the most valuable thing in the report. It's a direct readout of what your customers want to know and your website doesn't say.
What this looks like on a calendar
| When | What happens | Your effort |
|---|---|---|
| Day 0 | You send your URL | 2 minutes |
| Day 0 | We crawl your site and build a working bot | none |
| Day 0–1 | We show it answering your real questions | 15-minute call |
| Day 1 | You approve the content list and branding | ~15 minutes |
| Day 1 | You paste one line into your site | 5 minutes |
| Weekly | We re-read your site | none |
| Monthly | You get the questions report | reading it |
What we need from you
| You provide | Why |
|---|---|
| Your website URL | It's the bot's knowledge |
| Extra documents (optional) | Anything not already on the site |
| Logo and brand colours | So the widget looks like yours |
| A contact channel | Where the bot sends people it can't help |
| Your production domains | We lock the widget to them |
That's it. No integrations, no access to your systems, no data export from your CRM.