Creating Knowledge

The Knowledge Base transforms your documents, tables, and reference materials into an intelligent, searchable layer that powers AI agents, workflows, and dashboards. This guide walks you through creating and organizing knowledge effectively.

What Counts as Knowledge?

Knowledge refers to any material that supports internal decision-making or automation:

• SOPs and how-to guides

• Reference tables (thresholds, pricing, validation rules)

• Draft templates or sample phrasing

• Connected data from external systems (CRMs, databases)

• Synced content from Notion, Google Drive, SharePoint, etc.

Once added, this content can be queried by agents, referenced by workflows, or used in dashboards, without needing to manually search.

Three Methods for Creating Knowledge

1. Uploading Documents

Use this method for static files stored locally or when you need one-time uploads:

Best for:

• PDF, Word, or text documents

• CSV or Excel spreadsheets

• Policy documents, templates, archived reports

Steps:

1. Click "Add Source" in the Knowledge Base

2. Select Upload tab

3. Choose the file(s) you want to upload using drag-and-drop or file browser

4. Monitor upload progress with real-time indicators

5. Assign to a Collection

6. Click "Add to Knowledge Base" to complete

Supported file types: PDF, DOC, DOCX, TXT, CSV, XLSX

2. Connecting Database Tables

Use this for dynamic data that requires scheduled updates:

Best for:

• CRM records (Salesforce, HubSpot)

• ERP data (QuickBooks, SAP)

• Customer databases

• Sales performance data

Steps:

1. Click "Add Source" > Connect tab

2. Select your database integration

3. Provide API credentials or authentication

4. Choose specific tables, records, or data views to sync

5. Configure sync schedule (hourly, daily, weekly)

6. Assign to a Collection for organization

Sync behavior: Scheduled updates with status monitoring and error handling.

3. Connecting External Apps

Use this for live, collaborative content that changes frequently:

Best for:

• Team communication (Slack channels)

• Shared documents (Google Drive, Notion)

• Collaborative workspaces (SharePoint, Airtable)

Supported integrations include:

• Google Drive (documents, folders)

• Slack (channel content, message history)

• Notion (pages and databases)

• SharePoint (document libraries)

• Airtable (bases and tables)

Steps:

1. Click "Add Source" > Connect tab

2. Select the platform you want to connect

3. Authenticate your account with OAuth

4. Select which folders, channels, or databases to sync

5. Configure sync preferences (real-time vs. scheduled)

6. Assign to a Collection

Sync behavior: Real-time updates via webhooks or polling, depending on the platform's capabilities.

Organizing Knowledge with Collections

Collections help you group related knowledge by topic, function, or audience. Think of them as smart folders that workflows and agents can reference specifically.

Why Use Collections?

• Ensure agents only access relevant material

• Scope retrieval to avoid noise and irrelevant results

• Maintain separation by workflow, department, or use case

• Improve search accuracy and response quality

Collection Examples:

Collection NameContentsUsed ByCustomer Support DocsFAQs, troubleshooting guides, response templatesSupport agents, chatbotsFinance RulesApproval thresholds, policy documents, proceduresApproval workflows, finance dashboardsSales MaterialsPitch decks, product briefs, competitive analysisSales agents, proposal workflowsHR PoliciesEmployee handbook, benefits info, proceduresHR workflows, employee self-service

How to Create a Collection:

1. Go to the Collections tab in the Knowledge Base

2. Click "New Collection"

3. Name the collection descriptively

4. Add a clear description of its purpose and contents

5. Add existing sources or upload new content directly into the collection

6. Configure access permissions if needed

Advanced Knowledge Structuring

Structured Tables for Q&A

Use this approach when your knowledge comes in a question-answer or rule-based format:

Recommended structure:

• Question column: Common queries or scenarios

• Answer column: Detailed responses or guidance

• Category/Tag column: For filtering and organization

• Priority column: For handling multiple applicable answers

• Last Updated column: For content freshness tracking

Tips for effectiveness:

• Use consistent phrasing and terminology

• Include variations of common questions

• Add examples in separate columns when helpful

• Keep answers concise but complete

Chunking & Formatting Strategy

When processing long documents, the system breaks content into "chunks" for better indexing and retrieval:

Best practices:

• Use clear headers and subheaders to guide chunking

• Keep related information together in logical sections

• Avoid uploading massive files as single blocks

• Break complex procedures into step-by-step sections

Document structure tips:

• Use consistent heading hierarchy (H1, H2, H3)

• Include white space between sections

• Use bullet points and numbered lists appropriately

• Add table of contents for long documents

Metadata & Tags

Enhance content clarity and filterability with strategic metadata:

Recommended metadata fields:

• Status tags: "active," "archived," "draft," "under review"

• Version information: "v2.1," "2024-Q1," "latest"

• Department/owner: "HR," "Finance," "Sales," "IT"

• Content type: "policy," "template," "reference," "procedure"

• Audience: "managers," "all-staff," "customer-facing"

How to add metadata:

1. During upload, use the metadata fields in the source configuration

2. For connected sources, configure field mapping to capture relevant metadata

3. Bulk edit metadata for existing sources in the Collections view

Quality Management

Keep your knowledge base reliable and current:

Regular maintenance:

• Quarterly reviews: Audit content for accuracy and relevance

• Version control: Remove outdated files and update versioning

• Usage analysis: Monitor which knowledge gets accessed most frequently

• Owner validation: Have team owners review and approve new knowledge

• Dead link checking: Verify external references and links still work

Content quality checklist:

• Is the information current and accurate?

• Are file names and descriptions clear and searchable?

• Is the content properly tagged and categorized?

• Does the formatting support good chunking?

• Are there any duplicate or conflicting sources?

Connecting Knowledge to System Components

Once created and organized, your knowledge powers multiple parts of the system:

AI Agents

Purpose: Answering questions, summarizing content, providing context

To connect:

1. Open the Agent's configuration panel

2. Go to "Connected Resources" > "Knowledge"

3. Select one or more Collections

4. Configure retrieval settings:

   • Search method (semantic, hybrid, filtered)

   • Result limits and ranking

   • Fallback behavior for no matches

Workflows

Purpose: Decision logic, validation rules, routing criteria

To connect:

1. In workflow builder, add a "Knowledge Query" step

2. Select target Collections

3. Define query parameters and filters

4. Map results to workflow variables

Dashboards

Purpose: Reference data, rule mappings, threshold values

To connect:

1. In dashboard configuration, add a "Knowledge Reference" component

2. Select Collections containing reference data

3. Configure display format and update frequency

Best Practices Summary

For maximum effectiveness:

• Start with high-impact, frequently-used knowledge

• Use consistent naming conventions across all sources

• Organize Collections by usage pattern, not just topic

• Regularly validate knowledge accuracy with subject matter experts

• Monitor usage analytics to identify gaps or redundancies

• Test knowledge retrieval in realistic scenarios before deploying to production