Overview
The EKB Knowledge Base is where you store and manage content that agents can search and use — uploaded files, URL imports, web crawls, connector-backed documents, guidance (Q&A / templates), and Smart Tables.
Open it from the project sidebar: Knowledge Base.
Knowledge Base sidebar
The left sidebar is the main map of the Knowledge Base:
| Section | What it is |
|---|---|
| KB Data | Main file and folder browser for ingested documents |
| Guidance | Q&A pairs, response templates, and examples that steer agent answers (Templates) |
| Documents | Create and edit text documents in-product |
| Crawlers | Web crawlers that index sites into the Knowledge Base (Using Web Crawlers) |
| Smart Tables | Structured tables for extraction and automation (Smart Tables) |
Under Smart Tables, open any listed table to work in it, or use + New Smart Table to create one. See Smart Tables.
Chunking, connectors, and sync schedules are configured under Project Settings → Knowledge Base (not in this sidebar). See KB Settings, Connections, and Sync & Schedule.
KB Data
KB Data is the default view: a directory of folders and files with sync status, size, and actions.
Toolbar
| Control | Purpose |
|---|---|
| + New | Add content — upload files or folders, URL Upload, create a document, or start a Crawl Website flow |
| Storage / word count | Total size and word count for the Knowledge Base |
| Download All | Download Knowledge Base data as a ZIP |
| Search | Find items by name |
| All Types / All Status | Filter the list by file type and processing status |
| Create Folder | Add a folder in the current directory |
Directory list
The table shows Main Directory (or the folder you opened) with columns such as Name, Date modified, Added by, Size, and Actions.
- Folders and files show a status badge (for example synced) under the name when processing is complete
- Actions (⋯) opens item options such as Sync, Delete, and related resource actions
Common ways to add content:
- Upload files
- Add Confluence content or Add Google Drive content (after the connector is connected)
- Using Web Crawlers
Document detail view
Opening a file shows the document viewer: content tabs plus a details sidebar.
Content area tabs
| Tab | Shows |
|---|---|
| View | The file as uploaded (when a preview is available) |
| Content | Extracted text used for retrieval |
| Chunks | Segments produced by the project’s chunking settings |
Details sidebar (high level)
| Area | Purpose |
|---|---|
| File details | Name, upload info, size, type, character count |
| Document data / Smart Table | Attach a Smart Table schema for structured extraction |
| Custom Metadata | Descriptive fields — User-Defined (reference) vs AI-Searchable (retrieval) |
| Access Tags | Who can see the document (Allow Only / Allow None) |
Full guide: Tags & Custom Metadata.
Associated questions on a document can capture common Q&As tied to that file’s content.
Guidance, Documents, Crawlers, Smart Tables
| Area | Start here |
|---|---|
| Guidance | Knowledge Base Templates |
| Documents | Create documents from the Documents sidebar item (or + New → create document from KB Data) |
| Crawlers | Using Web Crawlers |
| Smart Tables | Smart Tables — + New Smart Table in the sidebar |
Troubleshooting
| Symptom | What to check |
|---|---|
| New files not searchable | Wait for status to reach synced; open the file → Content / Chunks; use ⋯ → Sync to reprocess |
| Connector content missing | Confirm the connection under Connections; re-add or sync the resource; see Add Confluence Content or Add Google Drive Content |
| List looks empty | Clear All Types / All Status; check you’re not inside a nested folder; use Search |
| Document hidden for some users | Review Access Tags on the file (Tags & Custom Metadata) |
| Chunking / retrieval quality | Review KB Settings (chunk size, strategy, enrichment), then re-sync affected documents |
FAQ
Can the Knowledge Base be partitioned by entity?
Yes. Knowledge Bases can be partitioned using Projects, so content is organized and isolated by the relevant entity (for example, a customer, business unit, or product line).
What preparation is required before ingestion?
Organize data into the appropriate, searchable folders before ingestion. Clear folder structure helps ensure content is set up correctly for retrieval.
What is the main latency advantage of moving from GCP Vector stores to Enterprise KB?
We are validating this internally and will publish the relevant details when they are available.
Is the retrieval trace stored and auditable later?
Yes. Vector chunks are stored within the respective Projects in the vector database, so retrieved content can be audited later.
How do we measure and improve retrieval quality?
The UI does not currently expose a dedicated retrieval-quality metric. Relevant retrieval information can be obtained through the API for further analysis.
Are there size or volume limits?
There are no specific limits based on Knowledge Base size or volume. Credit consumption scales with usage — see Credits.