Skip to main content

Getting Started

In about ten minutes, you'll have a MimicBot chat widget running on your site, answering visitor questions from your crawled content — with inline citations pointing back at the exact pages the answer came from.

What MimicBot does in one paragraph

MimicBot crawls the public pages of your site, embeds each chunk, and serves a floating chat widget that answers visitor questions by retrieving the most relevant chunks and generating a grounded response with citations. When you enable actions, the same widget can also complete contact forms, trigger lead captures, and other structured side-effects on behalf of your visitors — still grounded in what your site actually says.

The five steps

  1. Install the widget on your site with a single script tag. No bundler plugin, no framework integration — just a <script> element and the widget boots into a shadow DOM so it cannot collide with your page CSS.
  2. Create a bot via the REST API. One POST /api/bots call seeds the bot record, the widget config, and kicks off the first crawl.
  3. Watch the first crawl complete automatically. Crawl status moves from queuedrunningcompleted; pages land in Postgres as they're scraped and embedded.
  4. Verify the installation is detected. The dashboard marks the bot as "installed" once it sees a GET /v1/public/bots/:publicId/config from a real browser session on your site.
  5. Open the widget and ask a question. Confirm the answer is grounded in your content and that the citation links resolve back to real pages.

Each step has its own page below. Start with Install the widget.

What you'll need

  • A site you control (static HTML, WordPress, Next.js, anything — MimicBot is shadow-DOM isolated so it will not clash with your existing styles or scripts).
  • A MimicBot agency account with at least one API token. Tokens live under Settings → API tokens and are scoped to a single agency.
  • Five minutes of wall-clock attention while the first crawl runs. Small sites finish in under a minute; larger ones may take a couple of minutes depending on the number of pages and your configured crawl cap.

How long each step really takes

StepTypical timeWhat blocks it
Install the widget30 secondsPasting the script tag into your layout template
Create a bot30 secondsOne curl call
First crawl30 seconds – 3 minutesSite size and maxPages cap
Verify installation10 secondsOne full page load after the script lands
Ask a question5 secondsLLM response latency

If the crawl is still running after a few minutes on a small site, check the crawl status page for partial results — you can start asking questions as soon as the first page lands.

Next

→ Install the widget