Index Knowledge Tool

The Index Knowledge Tool enables AI agents to autonomously capture and preserve valuable knowledge they discover during operations. Rather than letting insights from research, email analysis, or SME consultations disappear after execution, agents can index this knowledge directly into your Corporate Knowledge Index for future retrieval.

Overview

The Problem

When agents process emails, conduct research, or synthesize information from multiple sources, they often uncover valuable insights that would benefit the organization. Without the Index Knowledge Tool:

• Research findings from SME consultations are lost after the conversation ends
• Meeting summaries extracted from email threads vanish after processing
• Synthesized knowledge from multi-source analyses disappears with the execution context
• All knowledge entry requires human action, creating bottlenecks

The Solution

The Index Knowledge Tool gives agents the ability to store valuable knowledge directly into your Corporate Knowledge Index. When an agent determines that information has lasting value, it can:

1. Index the content with appropriate metadata
2. Auto-classify the topic using AI when metadata is not provided
3. Make the knowledge searchable for all authorized agents

This creates a continuously growing corporate knowledge base, enriched by your AI agents' daily operations.

Prerequisites

Before agents can use the Index Knowledge Tool, you need:

1. Corporate Knowledge Index - A provisioned Corporate Knowledge index (see Corporate Knowledge Setup)
2. Feature Enabled - The Index Knowledge Tool must be enabled in your tenant settings
3. Tool Assignment - Agents must have the index_knowledge tool assigned to use this capability

Enabling the Tool

Step 1: Navigate to Settings

1. Sign in to Control Bridge
2. Navigate to Build > Knowledge > Index Knowledge Tool

Step 2: Enable the Feature

The settings page displays an overview of your configuration and usage statistics.

1. Toggle Enable Index Knowledge Tool to ON
2. The configuration section expands to show additional options

Step 3: Select Target Index

Choose which Corporate Knowledge Index agents should index into:

1. From the Corporate Knowledge Index dropdown, select your target index
2. Only indexes with status "Active" appear in the dropdown
3. If no indexes appear, you need to create a Corporate Knowledge Index first

Step 4: Configure Rate Limits

Set limits to control agent indexing activity:

Setting	Default	Description
Max Documents per Hour	10	Maximum documents any single agent can index per hour
Max Documents per Day	50	Maximum documents any single agent can index per day
Max Total Documents	10,000	Maximum total agent-indexed documents across all agents
Max Content Size	100 KB	Maximum size of content per document

tip

Start with the defaults and adjust based on your agents' activity. You can monitor usage on this same page to identify if limits need adjustment.

Step 5: Save Settings

Click Save Changes to apply your configuration.

Assigning the Tool to Agents

After enabling the feature, you need to assign the index_knowledge tool to agents that should have indexing capability.

1. Navigate to Build > AI Agents > Agents
2. Edit the agent you want to enable
3. In the Tools section, select Index Knowledge from the available tools
4. Save the agent configuration

Recommended Agents

Consider enabling this tool for:

Agent Type	Use Case
Research SMEs	Index research findings, synthesized knowledge, and expert conclusions
Executive Assistant	Store meeting summaries, decision records, and project outcomes
AICOS	Capture organizational learning and strategic insights
Email Processing Agents	Index valuable email content identified during processing

How Agents Use the Tool

When an agent has the index_knowledge tool assigned, it can index content by providing:

Required Parameters

• content - The knowledge content to index (minimum 10 characters, maximum 100 KB)

Optional Parameters

Parameter	Description	Auto-Generated If Omitted
title	Title for the document	Generated from first 100 characters of content
summary	Brief summary	LLM-generated summary
source	Origin of the knowledge (e.g., "Email thread analysis")	Defaults to agent name
documentType	Type: research, meeting_notes, decision, procedure, reference, other	No
topicId / topicName	Topic from taxonomy	Auto-classified using AI
keywords	Keywords for search optimization (max 10)	Auto-extracted from content
valueScore	Importance score (0.0-1.0)	Defaults to 0.8

Auto-Classification

When agents omit topic or keyword metadata, the system automatically:

1. Analyzes the content using the Topic Classification Model
2. Matches to the most appropriate topic in your Topic Taxonomy
3. Extracts relevant keywords for search optimization
4. Records both provided and generated metadata for transparency

Example Agent Prompt

To guide your agents on when to use this tool, consider adding instructions like:

When you discover valuable information that could benefit the organization:
1. Assess whether the information has lasting value beyond this execution
2. If valuable, use the index_knowledge tool to preserve it
3. Provide clear titles and appropriate document types
4. Let the system auto-classify topics if you're unsure

Good candidates for indexing:
- Research findings that answer common questions
- Policy clarifications from authoritative sources
- Process documentation extracted from email threads
- Decision records and their rationale

Monitoring Usage

Settings Page Statistics

The Index Knowledge Tool settings page displays real-time usage statistics:

Metric	Description
Total Indexed	Total documents indexed by all agents
Last Hour	Documents indexed in the past hour
Last 24 Hours	Documents indexed in the past 24 hours
Duplicates Blocked	Documents rejected as duplicates
Errors	Failed indexing attempts
Percent Used	Percentage of total document quota used

Top Indexing Agents

The settings page also shows which agents are most active in contributing knowledge:

• Agent name and friendly name
• Total documents indexed
• Last indexing timestamp

Detailed Logs

For detailed analysis, navigate to Manage > Data & Logs > Index Knowledge Logs.

The logs page provides:

Filters

• Agent - Filter by specific agent
• Outcome - Filter by result: indexed, duplicate, error, quota_exceeded
• Date Range - Filter by start and end dates

Statistics Cards

• Total Indexed - Successful indexing operations
• Classification Rate - Percentage using auto-classification
• Duplicates Blocked - Duplicate content rejected
• Errors - Failed operations

Log Details

Click any log entry to view detailed information:

• Document title and ID
• Content size
• Metadata provided by agent
• Metadata generated by system
• Outcome and any error message
• Timestamp

Duplicate Prevention

The Index Knowledge Tool uses SHA-256 content hashing to prevent duplicate documents:

1. When an agent attempts to index content, a hash is computed
2. The system checks if this hash already exists in the index
3. If a duplicate is found, the indexing is blocked
4. The log records this as a "duplicate" outcome

This ensures your knowledge base remains clean and agents don't waste quota on redundant content.

Enabling During GKI Creation

When creating a new Corporate Knowledge Index using the Grounded Knowledge Index wizard, you can enable AI Knowledge Indexing as part of the setup process.

On Step 5 (Configure Your Index), when creating a Corporate Knowledge index:

1. A toggle appears: Enable AI Knowledge Indexing
2. Enable this option to automatically:
   • Enable the Index Knowledge Tool for your tenant
   • Set this new index as the target for agent indexing
   • Apply default rate limits

This streamlines the setup process so you can start indexing immediately after provisioning completes.

Troubleshooting

Agent Cannot Index

Symptoms: Agent calls fail with "feature not enabled" error

Solutions:

1. Verify the Index Knowledge Tool is enabled in settings
2. Confirm the agent has the index_knowledge tool assigned
3. Check that a Corporate Knowledge Index is selected

Quota Exceeded Errors

Symptoms: Indexing fails with "quota exceeded" message

Solutions:

1. Check the usage statistics on the settings page
2. Identify which agents are consuming the most quota
3. Consider increasing limits if the indexing is valuable
4. Review agent instructions to reduce unnecessary indexing

Duplicate Rejections

Symptoms: Many "duplicate" outcomes in logs

Solutions:

1. This is expected behavior for duplicate content
2. If legitimate updates are being rejected, content must differ
3. Consider having agents append timestamps or version info for updated content

Auto-Classification Failures

Symptoms: Documents indexed without topic classification

Solutions:

1. Verify the Topic Classification Model is configured in Account Settings
2. Check that the target index has a Topic Taxonomy configured
3. Review the fallback behavior (documents are indexed as "Uncategorized")

Index Status Not Active

Symptoms: No indexes appear in the dropdown

Solutions:

1. Navigate to Build > Knowledge > Corporate Knowledge
2. Verify you have at least one Corporate Knowledge index
3. Check the index status is "Active"
4. If "Error" status, resolve the provisioning issue first

Security Considerations

The Index Knowledge Tool includes several security measures:

Measure	Description
Tenant Isolation	Agents can only index to their tenant's designated index
No Arbitrary Indexing	Cannot specify target index (always uses designated index)
Content Size Limits	Maximum content size prevents abuse
Rate Limiting	Per-agent hourly/daily limits prevent overwhelming
Audit Trail	All indexing attempts are logged for compliance
Tool Assignment	Only agents with explicit tool access can index

Best Practices

Content Quality

• Encourage agents to provide descriptive titles
• Use appropriate document types for better organization
• Include source attribution for traceability
• Set value scores based on content importance

Rate Limit Tuning

• Start with defaults and monitor for two weeks
• Increase limits for high-value agents that produce quality content
• Decrease limits for agents that generate low-value duplicates
• Use daily limits to prevent runaway indexing

Agent Instructions

• Provide clear guidance on what merits indexing
• Encourage quality over quantity
• Guide appropriate use of document types
• Remind agents to check for existing knowledge before creating duplicates

Monitoring

• Review logs weekly for patterns
• Track classification rate to understand metadata quality
• Monitor error rates for system issues
• Check quota usage to plan for growth

Related Topics

• Corporate Knowledge Setup
• Grounded Knowledge Indexes
• Topic Taxonomy
• Tools
• Agents
• AI Chief of Staff