Skip to main content

Update Your Knowledge Base Without Leaving Chat

Instead of manually editing QA pairs in your Library, you can update them conversationally through Chat. Simply describe what changed, and the AI will find and update the relevant content for you.

How It Works

1

Describe the Change

Tell the AI what information has changed in natural language
2

AI Finds Relevant Content

The system searches your Library and identifies QA pairs that need updating
3

Review Proposed Changes

See exactly what will change with word-level diff highlighting
4

Accept or Modify

Approve changes individually or in bulk, or edit before applying

Example

You say: “We now have 120 employees instead of 100.” AI responds: I found 3 QA pairs mentioning your employee count:
  1. Company Overview - “…our team of 100 employees…” → “…our team of 120 employees…”
  2. About Us - “100 dedicated professionals…” → “120 dedicated professionals…”
  3. Team Size - “Currently 100 employees…” → “Currently 120 employees…”
Would you like me to apply these changes?

Key Features

Intelligent Matching

The AI doesn’t just do text replacement - it understands context:
  • Finds all variations of how information appears
  • Identifies related content that might need updates
  • Considers synonyms and different phrasings
  • Preserves the original tone and style of each document

Word-Level Diff View

See exactly what will change before you approve:
  • Additions highlighted in green
  • Removals highlighted in red
  • Side-by-side comparison of old vs new
  • Full context preserved around changes

Smart Response Generation

When updating content, the AI:
  • Corrects typos in surrounding text
  • Matches the tone of the original document
  • Integrates changes naturally into existing sentences
  • Maintains consistent formatting

Common Update Scenarios

Updating Facts & Figures

  • “Our revenue is now 10M,not10M, not 8M”
  • “We moved offices to 500 Main Street”
  • “Our CEO is now Sarah Chen”

Updating Product Information

  • “We added support for Python 3.12”
  • “Our API rate limit increased to 1000 requests/minute”
  • “We now offer 99.99% uptime SLA instead of 99.9%“

Updating Compliance Status

  • “We completed our SOC 2 Type II audit”
  • “Our ISO 27001 certification was renewed”
  • “We’re now GDPR compliant”

Review & Approval Workflow

Individual Review

For each proposed change, you can:
  • Accept - Apply the change immediately
  • Reject - Skip this change, keep original
  • Edit - Modify the proposed text before applying

Bulk Operations

When many QA pairs need the same update:
  • Accept All - Apply all proposed changes at once
  • Reject All - Cancel all proposed changes
  • Select specific changes to apply or reject

Auto-Approval

When you accept a change:
  1. The QA pair is updated with your approved text
  2. The QA pair is automatically marked as approved
  3. It’s immediately available for response generation

Best Practices

Instead of “update our company info,” say “Our employee count changed from 100 to 120.”
The AI might find content you didn’t know existed. Review to ensure all changes are appropriate.
For clarity, make updates in focused batches rather than combining multiple unrelated changes.
Quarterly metrics, annual certifications, and regular updates are perfect for this workflow.

Permissions

To use knowledge updates via Chat:
  • You must have Edit access to the source containing the QA pairs
  • Private sources require explicit edit permission
  • Public sources can be edited by any organization member