Knowledge
One library. Your team quotes it, and so does Scout.
A passage becomes quotable when a named person approves it. Scout cites the passage and its approval date in the thread, agents see the same text beside the conversation, and customers read it in the Help Center. Three surfaces, one source.

Illustrative demo with fictional data.
One source
Write the answer once. It shows up in three places.
The same approved FAQ answers a customer on WhatsApp through Scout, sits beside an agent as a suggestion in the inbox, and publishes as a Help Center article. Edit the source and all three change together.
The customer finds it first
Publish the same answer as a Help Center article so the question is handled before it becomes a conversation. Published articles sync back as knowledge Scout can quote.
Approved knowledge item
FAQReturns and refunds
Items can be returned within 30 days of delivery if unused and in their original packaging.
Illustrative demo with fictional data.
What goes in
Bring in what your team already writes.
Start with the questions that actually fill your queue, then add the policies behind them. Every item is split into passages, and a passage is the unit Scout retrieves and cites.
- FAQ
- A question and its answer. The fastest way to cover what customers ask most.
- Article
- A longer piece with a title and body, for guides and explanations that a single answer cannot carry.
- Text or Markdown
- Paste a policy, a price list or an internal note and save it as a document.
- Help Center sync
- Pull in published articles. The result tells you exactly how many were added, updated and archived.
- Collections and tags
- Group by collection, tag for filtering, and scope which assistant may use which collection.
Scout reads FAQs, articles, pasted text or Markdown, and published Help Center articles. PDF and DOCX uploads and website crawling are not supported — we would rather say so than index something badly.
Channels
WhatsAppInstagramEmailLive chatTeams
SupportOrdersKnowledge
Add FAQAdd articleAdd text or MarkdownSync Help CenterAdd FAQIllustrative demo with fictional data.
Approval
Nothing is quotable until a person approves it.
Knowledge is content you manage, not a background process. Approval is an act by a named person, it is dated, and Scout shows both when it quotes the passage.
- Held for review
- Text that reads like instructions aimed at an AI is held automatically. Scout will not retrieve it until someone approves it, which closes the obvious prompt-injection route into your knowledge.
- Approve, edit, archive
- Edit a passage, archive it, restore it or delete it. An archived item stops being retrievable the moment you archive it.
- Indexing you can see
- Each item shows its indexing state and passage count. Reindex one item or retry a failed batch, without a support ticket.
- Dated approvals
- The citation Scout posts in the thread carries the approval date, so an agent can tell a six-week-old policy from today's.
Illustrative demo with fictional data.
Retrieval
No approved passage, no sentence.
Scout searches the library for the question in front of it and scores what comes back. Above the threshold it answers and cites the passage it used. Below it, there is no clever fallback: the conversation goes to a person, and the gap shows up as an answer you still need to write.
- Search always matches on keywords. Vector similarity is added on top only when an embedding model is configured for your installation.
- Wording matters. A question phrased far from your FAQ — in Arabic especially — can still fall under the threshold, which is why you test with real questions before you ship.
- A handover is a result, not a failure. The queue of handed-over questions is the shortest list of answers worth writing next.
Test a customer question
Handed to a person
Nothing above the threshold. Scout would hand this to a person instead of guessing.
Illustrative matches on fictional knowledge.
Help Center
Publish the answers customers would rather read than ask for.
Run a portal per brand or audience, in English, Arabic or both. Articles move from draft to published to archived, visitors search them, and published articles sync back into the library Scout quotes.
- Portals and locales
- More than one portal, each published in the languages you choose, from the same workspace.
- Categories
- Organise articles so customers can browse as well as search.
- Draft, publish, archive
- Review before anything is public, and archive what has gone stale instead of leaving it to be quoted.
Channels
WhatsAppInstagramEmailLive chatTeams
SupportOrdersArticles
Nile Home Help CenterEnglish (default) · العربية
- How to track your orderOrdersENPublished
- Returns and refundsPoliciesENPublished
- الإرجاع واسترداد المبالغPoliciesARPublished
- Delivery timesDeliveryENDraft
- Old size guideProductsENArchived
Categories
- Orders
- Policies
- Delivery
- Products
Illustrative demo with fictional data.
Keep exploring
Knowledge is what the rest of it reads.
Questions
Knowledge questions, answered.
What can I add as knowledge?
FAQs, articles, pasted text or Markdown, and published Help Center articles. PDF and DOCX uploads and website crawling are not supported.
Will Scout ever answer from something nobody approved?
No. Retrieval runs over approved passages only. Text that reads like instructions aimed at an AI is held for review before it is retrievable, and when nothing clears the threshold Scout hands the conversation to a person.
Can agents use the same knowledge?
Yes. When a conversation's inbox has a Scout assistant, agents get the Copilot panel, which drafts from the same passages. Nothing is sent until the agent sends it.
Does search understand meaning, or only keywords?
Keyword matching always runs. Meaning-based vector search is layered on top when an embedding model is configured for your installation, which is a deployment decision rather than a plan feature.
Is the Help Center available in Arabic?
Yes. A portal supports English and Arabic locales, with categories and articles in each, and the public pages render right to left in Arabic.
Bring your customer conversations together.
Tell us a little about your team. We'll show you how BlendDesk could fit the way you already talk to customers.