Add a website as a Pulse knowledge source

Source types

Pulse pulls from four kinds of source. Mix and match: most shops use website + a few PDFs.

• Website: crawled on a schedule, respects robots.txt and sitemaps.
• PDF / DOCX upload: manuals, warranties, internal SOPs.
• Manual entry: quick Q&A pairs you write directly.
• URL list: a flat list of pages to fetch (great for help-center articles).

Add a website

1. Open /knowledge.
2. Click Add source → Website.
3. Paste the root URL (e.g. https://acmeplumbing.com).
4. Pulse auto-discovers your sitemap.xml. If you don't have one, it crawls links from the homepage.

Scope the crawl

Limit what Pulse indexes so visitor answers stay focused:

• Include patterns: /services/, /faq/, /pricing. Only crawl pages that match.
• Exclude patterns: /admin/, /careers/, /tag/*. Skip noisy or private sections.
• Max depth: 3. Avoid crawling pagination tails on blogs.

include:
  - /services/*
  - /service-area/*
  - /faq/*
exclude:
  - /admin/*
  - /careers/*
  - /blog/tag/*
maxDepth: 3

Refresh schedule

Pick a refresh cadence based on how often the site actually changes:

• Daily: high-traffic sites with constant updates.
• Weekly: most service shops. The default.
• Manual: set-it-and-forget-it pages, or you trigger from a CMS webhook.

Review what was indexed

1. Open /knowledge → click your source.
2. The page list shows everything indexed with last-fetched timestamps.
3. Click any page to see the chunks Pulse extracted.
4. If a page is missing important content, mark it Priority. Pulse weighs it more in retrieval.