Part 1 — Adding a Knowledge Source: Step-by-Step
The Knowledge Base is what PAM reads to answer your partners' questions. You can add Documents, Text, or Website sources. This guide walks through the full Website flow — the most powerful option, and the one with the most steps.
Step 1 — Open the Knowledge Source modal
Navigate to PAM AI → Knowledge and click the New Knowledge button in the top-right corner of the Knowledge Base table.
PAM AI → Knowledge — click "New Knowledge" to begin
Step 2 — Name your source and choose Website
Give your knowledge source a clear, descriptive name (e.g., "Euler Help Center" or "Partner Program Guide"). Then select Website as the Knowledge Type.
1. Enter a name 2. Select Website as the knowledge type
Step 3 — Add a URL
Type or paste your URL into the URL link field and click Save link. The URL appears as a card below the field, defaulting to Single page mode — PAM will only read that one page, with no discovery of other links.
URL saved in Single page mode (Map site is OFF by default)
Single page mode is a low-cost operation. Only enable Map site when you genuinely need to discover multiple pages from a URL. |
Step 4 — Enable Map site to discover links
If you want PAM to discover all the pages available under a URL, toggle Map site to ON on the URL card. The card turns blue and a hint confirms that links will be discovered.
A Discover links button then appears. Click it to start the mapping process.
1. Toggle Map site ON 2.Click Discover links to start
Step 5 — Review Map results and select links
Once mapping completes, a table of discovered links appears. Each row shows the page title, URL path, a Collection toggle, and a Scrape action.
Map results — 1. Summary bar 2. Collection toggle 3. Scrape button
What is Collection?
The Collection toggle (Yes / No) tells PAM whether a link is a parent/index page that contains links to sub-pages, or an individual content page.
Setting | When to use it | Effect on Scrape |
Yes
| Page is a section index — mostly links to other articles | Scrape button is enabled → click to fetch child pages |
No
| Page is a standalone article with actual content | Scrape button is disabled — page is indexed as-is |
Check the links you want PAM to index. Use Select all or pick individually. Only checked links are saved.
Step 6 — Try scraping if Map results are incomplete
If the pages you need aren't appearing in the map results, a hint appears at the bottom of the list:
"Not finding what you need?" — click Try scraping to run a second discovery method
Click Try scraping. PAM runs a different discovery method — it loads the page and extracts all visible hyperlinks, often returning results the map missed.
Once complete, a tab switcher appears at the top of the results panel, letting you switch between Map results and Scrape results. Each tab has its own independent selection.
1. Tab switcher 2. Collection = Yes (parent row) 3. Indented child pages discovered by scraping
Selections are independent per tab. When you save, only the links selected in the currently active tab are saved. To include links from both methods, save them as two separate knowledge sources. |
Step 7 — Save your selection
When you're happy with your selection, click Save selected. The button shows the count of selected links (e.g., "Save 5 selected"). Saved sources are added to your Knowledge Base as Active by default.
Managing your Knowledge Base
After saving, your sources appear in the PAM AI → Knowledge list. Each entry has a Status toggle and a Category label.
Status (Active/Inactive) and Category columns in the Knowledge Base
Active — PAM can read this source and use it in answers
Inactive — PAM ignores this source (useful for pausing outdated content without deleting it)
Collection — an index page that groups child pages under it
Website — an individual content page indexed directly
Part 2 — Understanding Map vs. Scrape
Map and Scrape are the two methods PAM uses to discover links from a URL. They work differently, return different results, and are best used in combination rather than as alternatives.
Map — Structure-based discovery
Map reads the site's sitemap or link graph to enumerate all declared pages under a given URL. It does not load the full content of each page — it only reads the structural declaration of what exists.
Best for: Well-structured sites with a public sitemap (help centers, docs portals, marketing sites)
What you get: A clean, comprehensive list of URLs reflecting the site's official hierarchy
What it may miss: Pages not in the sitemap — dynamically generated pages, recently published articles, or intentionally excluded pages
Scrape — Content-based discovery
Scrape loads the page and extracts all hyperlinks found in the rendered HTML — navigation menus, inline links, footer links. It reads what is actually visible on the page at scrape time.
Best for: Sites without a public sitemap, or when you need to fill gaps left by Map
What you get: Links as they appear on the live page — often includes pages the map missed
What it may miss: Pages only reachable via search, filters, or deep pagination
Side effect: May include navigation/footer links that aren't content pages — review results before saving
How to use them together
The recommended workflow is:
Start with Map — it's faster and gives you the broad structure
Select the pages you want
If you notice gaps, use Try scraping at the bottom of the results panel
Compare Map results and Scrape results using the tab switcher
Save from whichever tab best covers your needs
Depth limits when scraping Collections
When you scrape a Collection from within the results table, its child pages appear indented below. You can mark a child as a Collection and scrape it again. PAM supports up to 3 levels of depth:
Level | What it contains | Scrape available? |
Level 0
| Root links returned by Map or Scrape | Yes — if marked as Collection |
Level 1
| Children discovered from a Level 0 Collection | Yes — if marked as Collection |
Level 2
| Children discovered from a Level 1 Collection | No — Max depth reached
|
At Level 2, the Scrape button is replaced with a Max depth indicator. To access deeper pages, add that specific URL as a separate knowledge source. |
Quick reference
| Map | Scrape |
Method | Reads sitemap / link graph | Loads page, extracts live links |
Speed | Faster | Slightly slower |
Best for | Well-structured sites broad discovery | Filling gaps, dynamic or unsitemapped sites |
May miss | Pages excluded from sitemap | Pages behind search or deep pagination |
Result volume | High (full site structure) | Medium (links visible on a specific page) |
Troubleshooting
Map returned no results or very few links
The site may not have a public sitemap, or your URL is too specific (a leaf page rather than a section root). Try scraping instead, or use a higher-level URL such as the section index.
Scrape returned irrelevant links (navigation, footer)
This is expected — scrape extracts all visible links. Use the Selected tab to review picks before saving, and deselect anything that isn't relevant content.
A specific page isn't in either Map or Scrape results
Add that exact URL as a separate knowledge source without enabling Map site. It will be indexed as a single page directly.
I need content from both Map and Scrape
Save your Map selections first, then create a second knowledge source from the same URL and use Scrape. Each becomes an independent, separately manageable entry in your Knowledge Base.
Still Have Questions?
Reach out to our support team by:
Sending us a message via the in-app chat
Emailing us at [email protected]
We’ll get back to you as soon as possible!