Knowledge Base — Dashboard Access
Tenant admins access their Knowledge Base from Dashboard → Settings → Knowledge Base (/settings/knowledge-base). Datasets are created and configured by super admins in the Manage app — tenants interact with the datasets that already exist for their organization.
Full screen specs and flows: Manage — Knowledge Base Screens | Manage — Knowledge Base Flows
What Tenants Can Do
| Action | Screen | Can do? |
|---|---|---|
| View all their datasets (file counts, chunk counts, status) | KB1 — Overview | Yes |
| Upload documents to a dataset | KB2 — File Management + KB3 — Upload Modal | Yes |
| Remove documents from a dataset | KB2 — File Management | Yes |
| Enable / disable individual files (without deleting) | KB2 — File Management | Yes |
| Trigger a website re-crawl | KB6 — Crawl Settings modal | Yes |
| Test retrieval quality (sandbox) | KB4 — Retrieval Sandbox | Yes |
| Create a new dataset | KB5 — Dataset Configuration | No — Manage app only |
| Change embedding model | KB5 — Dataset Configuration | No — Manage app only |
| Change chunking strategy | KB5 — Dataset Configuration | No — Manage app only |
| Delete a dataset | KB5 — Dataset Configuration | No — Manage app only |
Where It Sits in the Dashboard
Knowledge Base is a sub-section of Settings, not a top-level nav item:
Settings
├── General
├── Knowledge Base /settings/knowledge-base ← here
│ ├── [datasetId] /settings/knowledge-base/[datasetId]
│ │ ├── Files tab
│ │ ├── Sandbox tab
│ │ └── (Configure tab — read-only view only)
│ └── Test Retrieval /settings/knowledge-base/[datasetId]/sandbox
├── Channels
├── Integrations
├── Skills
├── Recurring Tasks
├── Audit Log
└── BillingDatasets Available
Each tenant has up to four standard datasets (created by super admin on tenant setup):
| Dataset | What it holds | Who populates |
|---|---|---|
| Client Documents | Brand guides, tone-of-voice docs, product sheets | Tenant uploads |
| Website Content | Crawled pages from the client’s own website | Crawler (automated) |
| Published Content | Blog posts + social posts indexed after publishing | Platform (auto) |
| Competitor Research | Competitor pages scraped by the Competitor Researcher | Agent (auto) — local only |
Retrieved Context (in Activity Detail)
When an agent uses rag_search during a task, the activity card in Dashboard → Campaign Detail → Activity tab shows a collapsible “Retrieved context” section:
🤖 Copywriter — Writing blog post: "Why Local SEO Matters"
Status: ✅ Completed
▸ Retrieved context (3 queries)
Query 1: "local SEO for small businesses client services"
Dataset: client_docs · 3 chunks · Scores: 0.91, 0.87, 0.82
↳ brand-guidelines.pdf — "Our primary audience is local businesses…"
Query 2: "past blog posts about SEO"
Dataset: published_content · 3 chunks · Scores: 0.89, 0.85, 0.78
↳ "How to Rank on Google in 2026" — "When writing about SEO topics…"This lets tenants see exactly what context the agent used and identify when a re-upload or re-crawl would improve results.