Knowledge

Knowledge is what makes your AI agents intelligent and contextually aware. By adding knowledge from your Library, you equip agents with the information they need to provide accurate, relevant, and helpful responses. This guide covers how to add, manage, and optimize knowledge for your agents.

What is Agent Knowledge?

Agent knowledge refers to the information from your Library that you make available to an agent. When you add knowledge to an agent, that information becomes part of the agent's context - it's always "aware" of this information and can reference it during conversations.

Knowledge vs Tools:

Knowledge: Static information the agent knows (products, case studies, playbooks)
Tools: Actions the agent can perform (fetch CRM data, send emails)

Think of knowledge as the agent's "education" and tools as its "skills". Both are essential for an effective agent.

Types of Knowledge

You can add several types of knowledge to your agents from the Library:

Products

Product knowledge helps agents understand what you offer and recommend the right solutions.

When to add products:

Agents need to recommend products based on customer needs
You want agents to answer questions about features and pricing
Agents should identify cross-sell or upsell opportunities
You need agents to compare product options

What agents can do with product knowledge:

"Based on your needs, I'd recommend our Enterprise plan which includes..."
"Our Advanced Analytics module costs $X per month and provides..."
"Since you're already using our Core platform, you might benefit from adding..."
"Compared to the Standard tier, Enterprise includes these additional features..."

ICPs (Ideal Customer Profiles)

ICP knowledge helps agents qualify leads and tailor messaging to specific customer segments.

When to add ICPs:

Agents need to qualify inbound leads
You want agents to identify which products fit which customers
Agents should recognize expansion opportunities in accounts
You need agents to prioritize prospects based on fit

What agents can do with ICP knowledge:

"This prospect matches our Enterprise ICP - they're in the right industry and size range"
"Based on our ICP criteria, they appear to have strong buying authority"
"This account fits our expansion ICP for cross-selling Analytics"
"I notice this lead doesn't quite match our typical customer profile because..."

Case Studies

Case study knowledge provides social proof and demonstrates real-world value.

When to add case studies:

Agents need to overcome objections with proof points
You want agents to reference similar customer success stories
Agents should build credibility with prospects
You need agents to match prospects with relevant examples

What agents can do with case study knowledge:

"A similar company in your industry, [Customer X], saw a 40% reduction in costs"
"We have a great case study about [Company Y] who had the same challenge"
"Based on your use case, you might find this customer story relevant..."
"Other mid-market SaaS companies typically see ROI within 6 months, like [Customer Z]"

Playbooks

Playbook knowledge guides agents (and reps) through proven sales processes and methodologies.

When to add playbooks:

Agents need to suggest next steps in the sales process
You want agents to follow your established methodology
Agents should ensure consistent execution across the team
You need agents to provide coaching based on deal stage

What agents can do with playbook knowledge:

"According to our Enterprise playbook, the next step is to schedule a technical validation call"
"You should complete these checklist items before moving to the proposal stage"
"Our playbook suggests identifying economic buyer by this point in the cycle"
"Based on where you are in the deal, I recommend focusing on..."

Documents

Document knowledge provides reference material for specific questions and scenarios.

When to add documents:

Agents need to reference battle cards for competitive situations
You want agents to access technical documentation
Agents should have objection handling guides
You need agents to reference FAQs or compliance information

What agents can do with document knowledge:

"According to our competitive battle card, here's how we differentiate from [Competitor]"
"Our compliance documentation indicates that we support SOC 2 Type II..."
"For this objection, our handling guide suggests emphasizing..."
"The technical specs show that we integrate with..."

Adding Knowledge to Agents

To add knowledge to an agent:

Navigate to Settings → AI → Agents
Select the agent you want to configure
Click on the "Knowledge" tab
Click "Add Knowledge"
Select a category (Products, ICPs, Case Studies, Playbooks, Documents)
Choose specific items from your Library to add
Save the agent configuration

The knowledge is now available to the agent and will be included in its context during conversations.

Foreground vs Background Agent Knowledge

You can configure different knowledge for foreground and background agents:

Foreground agents (toggle off): Use the "Knowledge" tab to configure knowledge for interactive conversations
Background agents (toggle on): Use the "Internal Knowledge" section to configure knowledge for automated workflows

Background agents typically need more focused knowledge optimized for their specific tasks (e.g., specific playbooks for updating deal stages, relevant case studies for matching patterns).

Strategic Knowledge Selection

You don't need to add all your Library content to every agent. Strategic knowledge selection keeps agents focused and responses relevant.

By Agent Type

Account Agents:

All products (for expansion opportunities)
Expansion and upsell playbooks
Case studies from similar industries
Competitive battle cards
Executive-level documents

Deal Agents:

Products specific to the deal
Relevant playbook for the deal type (new logo, expansion, etc.)
ICPs to help qualify
Case studies that match the prospect's profile
Pricing and packaging documents

Contact/Lead Agents:

Top-performing products
Prospecting and qualification playbooks
All ICPs for qualification
High-impact case studies for credibility
Outreach templates and messaging guides

Customer Success Agents:

Products the customer owns or could expand to
Onboarding and adoption playbooks
Support documentation
Success stories from similar customers
Training resources

By Use Case

Discovery and Qualification:

ICPs for qualifying fit
Discovery playbooks
Product overview documents

Solution Selling:

Detailed product information
Relevant case studies
ROI calculators and value documents
Competitive battle cards

Objection Handling:

Case studies with proof points
Objection handling documents
Competitive differentiation documents

Closing:

Closing playbooks
Pricing and contract documents
Success stories with fast time-to-value

How Agents Use Knowledge

When you add knowledge to an agent, it becomes part of the agent's system context. Here's how agents access and use this knowledge:

Knowledge as Context

Knowledge items are converted to text and included in the agent's system prompt. For example:

Products: Name, description, features, pricing, use cases
ICPs: Company characteristics, roles, pain points, qualification criteria
Case Studies: Customer, challenge, solution, results, metrics
Playbooks: Stages, checklists, exit criteria, objectives
Documents: Full document content

Retrieval and Reference

Agents can:

Search knowledge: Find relevant information based on the conversation
Compare items: Contrast products, case studies, or playbooks
Apply methodologies: Follow playbook guidance step-by-step
Cite sources: Reference specific case studies or documents by name

Contextual Application

Agents use knowledge contextually:

When you ask about pricing, the agent references product knowledge
When qualifying a lead, the agent compares against ICP criteria
When facing objections, the agent pulls relevant case study metrics
When suggesting next steps, the agent follows playbook guidance

Knowledge Management Best Practices

Keep Knowledge Current

Regular reviews: Check knowledge items quarterly for accuracy
Immediate updates: Update agents when products, pricing, or processes change
Version awareness: Remove outdated case studies or deprecated products
Test after updates: Verify agents are using new information correctly

Be Selective

Quality over quantity: More knowledge isn't always better
Relevance matters: Only add knowledge that's relevant to the agent's purpose
Avoid overwhelming: Too much knowledge can dilute the agent's focus
Test and iterate: Add knowledge incrementally and test effectiveness

Organize Thoughtfully

Consistent naming: Use clear, descriptive names in your Library
Structured content: Well-organized knowledge is easier for agents to use
Related items: Link related products, ICPs, and case studies
Active only: Ensure only active Library items are added to agents

Complement with Tools

Knowledge and tools work together:

Knowledge provides the "what" (product info, methodology)
Tools provide the "how" (fetch data, send email, update CRM)
Together they create powerful, capable agents

Example: An agent uses ICP knowledge to qualify a lead, then uses the researchCompanyByDomain tool to gather additional firmographic data, and finally uses playbook knowledge to suggest the next step.

Updating Agent Knowledge

When you update knowledge in the Library, agents will see the changes, but there are some nuances:

When Changes Take Effect

New conversations: Start fresh to get updated knowledge
Active conversations: May still use cached context from when the conversation started
Background agents: Pick up changes on next execution

Best Practice for Updates

Update the content in the Library (Settings → Library)
Verify the update is saved
Start a new conversation with the agent to test
Ask a direct question about the updated content
Confirm the agent provides the new information

Testing Agent Knowledge

The agent testing interface is perfect for validating knowledge configuration:

Select a test object (Account, Contact, Deal, Lead)
Ask knowledge-specific questions:
- "What products would you recommend for this account?"
- "Does this prospect match any of our ICPs?"
- "Do you have any relevant case studies?"
- "What should be my next step according to our playbook?"
Verify the agent references the correct knowledge
Check for accuracy - is the information current and correct?
Test edge cases - what if the agent doesn't have relevant knowledge?

What to Test

Completeness: Does the agent have all necessary knowledge?
Accuracy: Is the information correct and up-to-date?
Relevance: Does the agent reference appropriate knowledge for context?
Gaps: Are there questions the agent can't answer that it should?

Common Knowledge Patterns

Product-ICP-Case Study Triangle

Link these three knowledge types for powerful selling:

Add products to the agent
Add ICPs that are good fits for those products
Add case studies from customers who match those ICPs

The agent can now recommend the right product, qualify against the right ICP, and reference the right case study - all contextually.

Playbook-Driven Knowledge

Build knowledge around playbooks:

Add a playbook (e.g., "Enterprise Sales Playbook")
Add products referenced in the playbook
Add case studies for proof points at specific stages
Add documents for templates and resources mentioned

The agent can guide reps through the entire playbook with all supporting materials.

Role-Based Knowledge

Configure different agents for different roles:

SDR Agent: Top ICPs, qualification playbooks, introductory case studies
AE Agent: All products, closing playbooks, comprehensive case studies
CSM Agent: Customer's products, adoption playbooks, success stories
Manager Agent: All playbooks, methodology documents, coaching resources

Troubleshooting

Agent not using knowledge:

Verify knowledge is added in the Knowledge tab
Check that Library items are set to "Active" status
Start a new conversation (old conversations use cached context)
Ask direct questions about the knowledge ("Tell me about [Product X]")
Review the system prompt to ensure knowledge is included

Agent providing wrong information:

Update the Library content with correct information
Verify you updated the right item
Start a new conversation to pick up changes
Check if multiple similar items exist and causing confusion
Ensure knowledge isn't conflicting (e.g., two products with similar descriptions)

Agent seems overwhelmed with too much knowledge:

Remove less relevant knowledge items
Create specialized agents for specific purposes
Break up large documents into focused pieces
Use more specific agent system prompts to guide knowledge use

Agent not finding the right knowledge:

Improve Library item names and descriptions
Add more context in product/ICP/case study details
Restructure knowledge for better discoverability
Guide the agent in the system prompt about when to use specific knowledge

Advanced: Knowledge in System Prompts

For advanced users, you can reference knowledge in your agent's system prompt to guide how it's used:

When recommending products, always consider our ICPs first. Match the prospect's profile against our ICP criteria, then suggest products that align with that ICP.

For enterprise deals, prioritize case studies from Fortune 500 companies. For mid-market deals, use case studies from companies with 100-1000 employees.

Always follow the playbook methodology. Don't suggest moving to the next stage unless the current stage's exit criteria are met.

This explicit guidance helps agents use knowledge more effectively and consistently.