Skip to main content

Open in AthenaHQ

app.athenahq.ai/knowledge-base

​
Purpose

The Knowledge Base is where a customer builds and maintains a structured “fact bank” about their business: the verified statements (claims) Athena should treat as ground truth when monitoring how AI search engines describe their brand. Claims are organized into pillars: themed groupings of related facts (e.g. “Pricing”, “Product Features”, “Company History”). This page sits at the center of the Oracle workflow. Athena uses the approved claims here to (a) detect when an AI assistant says something inaccurate about the customer’s business and (b) measure how well the customer’s own website content covers each pillar’s claims. Without a populated knowledge base, Oracle has nothing to compare AI responses against. Customers typically arrive here after onboarding to seed their fact bank: either by letting Athena auto-discover pillars from their site, by importing URLs and documents, or by pulling existing articles from Content Hub. They return regularly to review newly-extracted claims, manage pillars, and trigger fact verification runs. The page also tracks an internal link bank: pages on the customer’s own domain that Athena monitors as supporting content for pillar coverage.

​
What’s on the page

​
Page header

Shows the page title “Knowledge Base” and the primary action buttons in the top-right:
  • Verify facts. Opens a dialog where the user picks which pillars to verify against Oracle’s collected AI responses.
  • Add knowledge (dropdown). The main entry point for adding pillars and claims (see actions below).

​
Pillars tab

Lists every content pillar for the active website. Pillars are themed groupings of related claims and are the primary unit of organization in the knowledge base. Columns:
  • (checkbox). Select rows for bulk operations.
  • Pillar. The pillar name with optional description shown beneath in smaller text. Hovering the info icon in the header shows: “Content pillars that group related claims and measure content coverage.”
  • Content coverage. A horizontal bar showing the percentage of this pillar’s approved claims that are matched by existing content on the customer’s site. Hovering the bar reveals the exact percentage. Hovering the column header info icon shows: “Percentage of a pillar’s claims covered by your existing content.” This column is hidden on small screens (large breakpoint only). A dash (—) shows when no coverage data exists yet.
The Pillars tab trigger itself has a tooltip: “Content pillars with verified claims about your business.” Lists pages from the customer’s own domain that Athena tracks as part of the link bank: used for coverage analysis and as supporting context. Columns:
  • (checkbox). Select rows for bulk operations.
  • URL. The page URL. Clicking opens the page in a new tab.
  • Title. The page’s scraped title. Hidden on small/medium screens.
  • Status. A badge showing processing status (e.g. finished, scraped, pending, failed).
The Internal Links tab trigger has a tooltip: “Pages from your own domain not yet associated with a pillar.”

​
Bulk selection command bar

When one or more rows are selected (in either tab), a floating action bar appears at the bottom of the page showing:
  • N selected. Selection count.
  • Edit (pillars, single selection only, shortcut E). Opens the Edit Pillar dialog.
  • Delete (shortcut D). Removes the selected rows after confirmation.
  • Verify facts (pillars only, shortcut V). Opens the Verify Facts dialog pre-scoped to eligible selected pillars (published pillars with at least one claim).
  • Reset (shortcut Esc). Clears the selection.

​
What you can do here

​
From the page header

  • Verify facts. Opens the verification dialog. Choose which pillars to check; Athena cross-references their claims against AI responses Oracle has already collected and flags contradictions. After starting, the user is taken to the Oracle page.
  • Add knowledge (dropdown). Reveals five options grouped into two sections:
    • Pillars
      • Create pillar. Open a dialog to manually create a new pillar with a name, optional description, and optional “Auto-research” toggle that immediately starts research after creation.
      • Discover pillars. Opens a confirmation dialog. Athena analyzes the site to auto-discover relevant pillars and extract claims for each. Can take several minutes.
    • Knowledge
      • Add URLs. Opens a dialog to paste URLs or crawl a domain path. Toggles let the user add the URLs to the link bank, extract claims from them, or both. Shows the credit cost in the footer before submitting.
      • Upload document. Opens a dialog with two tabs: File (upload a PDF or DOCX, max 10MB) and Text (paste raw text, max 100,000 characters). Athena extracts claims for review.
      • Extract from Content Hub. Opens a dialog listing published articles from the customer’s Content Hub. Select articles and Athena extracts claims from them.

​
From the Pillars table

  • Click a pillar row. Opens the Pillar Detail dialog (see drilldowns).
  • Select rows + Edit. Opens an edit dialog to update the pillar’s name and description.
  • Select rows + Delete. Confirms then permanently deletes the pillar and all its claims. Confirmation reads: “Delete N pillar(s)? All claims for these pillars will be deleted.”
  • Select rows + Verify facts. Starts a verification run scoped to those pillars (only published pillars with ≥1 claim are eligible).
  • Click a URL. Opens the page in a new tab.
  • Select rows + Delete. Confirms then permanently removes those links from the link bank.

​
Inside the Pillar Detail dialog (drilldown)

Four tabs:
  • Claims. List of approved facts. Each row has a kebab (⋯) menu with a Delete option. Click a claim to see its full evidence in a sub-dialog.
  • Pages. Internal pages associated with this pillar. Each row has a delete button on hover. If empty, a Find matching pages button scans the site for pages whose content matches.
  • Document. Shows a synthesized markdown document built from the pillar’s claims. If none exists, a Generate document button creates one (requires at least one claim).
  • Matching articles. Existing site articles that cover this pillar’s claims, with a match-percentage badge and a count of matched claims per article.
The dialog header has an expand/collapse button (toggles between standard and full-screen), and a ⋯ menu with Edit and Delete actions for the pillar itself. At the bottom of the Claims tab, an Add knowledge dropdown offers:
  • Upload document. PDF/DOCX upload, claims extracted into this pillar.
  • Paste text. Pasted text, claims extracted into this pillar.
  • Import from URL. Opens the Add URLs dialog scoped to this pillar.
  • Extract from Content Hub. Pull and extract from Content Hub articles into this pillar.
  • Import CSV. Direct bulk-add of claims from a CSV file (no LLM extraction). Requires a header row with a column named claim (or text / fact); an optional source_url (or url / source) column is matched by name. A Download template link is provided.

​
Data shown

Everything on this page is scoped to the currently active website:
  • Pillars. The set of content pillars (sometimes called “topics”) the user has created, auto-discovered, or imported. Each pillar tracks a name, description, status, claim count, and last-researched timestamp.
  • Claims. Verified factual statements about the business, each attached to a pillar. Claims carry a source type (web research, AI response, upload, your site, manual), a confidence level (high/medium/low), and an optional source URL with quote.
  • Internal links. Pages on the customer’s own domain that Athena has crawled and indexed.
  • Content coverage. Computed by matching each approved claim against content already published on the site (typically Content Hub articles). The coverage percentage = (matched claims / total approved claims) per pillar.
  • Content Hub articles. When importing, the dialog lists the customer’s own published articles from Content Hub.

​
Common workflows

​
Build a knowledge base from scratch

  1. Click Add knowledge → Discover pillars.
  2. Confirm the Discover Pillars dialog. Athena analyzes the site, discovers relevant pillars, and starts extracting claims. This can take several minutes.
  3. Open each pillar (click the row) to review the extracted claims.
  4. When pillars look populated, click Verify facts to check those claims against the AI responses Oracle has collected.

​
Manually add a pillar and populate it

  1. Click Add knowledge → Create pillar.
  2. Enter a name, optional description, and optionally toggle Auto-research to start research immediately.
  3. Click Create.
  4. Click the new pillar row to open it.
  5. Use the Add knowledge dropdown inside the dialog to upload a document, paste text, import URLs, extract from Content Hub, or import a CSV.
  1. Click Add knowledge → Add URLs.
  2. Either paste URLs (one per line) or switch to the Crawl path tab and enter a domain/path; click Discover and pick which discovered URLs to import.
  3. Toggle Add to link bank and/or Extract claims depending on intent.
  4. Review the credit cost shown in the footer.
  5. Click Add N URLs.

​
Bootstrap claims from existing Content Hub articles

  1. Click Add knowledge → Extract from Content Hub.
  2. Select articles from the list (or use the select-all checkbox).
  3. Review the credit cost in the footer (1 credit per 10 articles).
  4. Click Extract from N article(s). Extraction runs in the background.

​
Bulk-delete stale pillars

  1. Check the boxes for pillars to remove. The floating command bar appears at the bottom.
  2. Click Delete (or press D).
  3. Confirm the destructive prompt.
  4. Selected pillars and all their claims are removed.

​
Empty, loading, and error states

  • Empty. When there are no pillars and no internal links yet, an icon + “No knowledge yet” message displays with the subtext “Discover pillars relevant to your website, or add knowledge manually.” and a Discover pillars button.
  • Loading. Skeleton placeholder rows and tab bar animate while data is being fetched.
  • Empty (per tab). If pillars exist but links don’t (or vice versa), the empty tab shows a simple “No pillars yet” / “No internal links yet” message.
  • Entitlement error. If the system can’t verify the customer’s plan access, a “Couldn’t verify access” screen with a Retry button appears.
Within dialogs:
  • File extraction shows an “Extracting text…” spinner.
  • Unsupported file types or files over 10MB show an inline error.
  • CSV parsing errors are shown inline (e.g. missing claim header, empty file).
  • Failed mutations surface a toast notification.
Linked from:
  • The main left sidebar, under the Manage section.
  • Direct URL /knowledge-base.
Links to:
  • Oracle (/oracle). After starting a fact-verification run, the user is automatically taken to Oracle to see results.
  • External URLs. Clicking a URL in the Internal Links table opens that page in a new tab. The same applies to source URLs in claim detail dialogs and matching articles.

​
Common support questions

Q: What’s a “pillar”? A pillar is a themed group of related claims about your business, like “Pricing”, “Product Features”, or “Customer Support”. Pillars are how Athena organizes the facts in your knowledge base and measures how well your site content covers each topic. Q: What does the “Content coverage” bar mean? It shows the percentage of that pillar’s approved claims that are already covered by content on your site. For example, 80% means 80% of the claims under that pillar are matched by existing pages. The exact percentage shows in a tooltip when you hover the bar. The column is only visible on larger screens. Q: Does adding URLs cost credits? Yes. Each batch of 10 URLs costs 1 credit (1–10 URLs = 1 credit, 11–20 = 2 credits, etc.). The exact cost is shown in the Add URLs dialog footer before you confirm. The same rate applies to Content Hub bootstrap (1 credit per 10 articles). Q: Why don’t my uploaded claims show up immediately? Claim extraction runs in the background. Once the dialog closes you’ll see a “Extraction started” toast: the claims appear in the pillar once the workflow finishes. CSV imports, by contrast, add claims directly without an extraction step, so they appear instantly. Q: Can I undo a pillar deletion? No. Deleting a pillar is permanent and also deletes all of its claims. The confirmation dialog warns of this. Q: What’s the difference between “Discover pillars” and “Create pillar”? Create pillar manually adds a single pillar with a name you choose. Discover pillars asks Athena to auto-analyze your site and create multiple relevant pillars for you, automatically researching each one and extracting claims. Discover takes several minutes; Create is instant. Q: What’s the “Verify facts” button for? It cross-references your approved claims against AI responses Oracle has already collected. If an AI assistant said something that contradicts a claim in your knowledge base, the verification run surfaces that discrepancy. After starting a run you’re taken to Oracle to see results. Q: What does the CSV upload format look like? A CSV with a header row. One column must be named claim (or text / fact). An optional column named source_url (or url / source) is matched by name: column order doesn’t matter. A Download template link inside the Import CSV tab gives you a starter file.