How to Integrate AI Agents with Your CRM

An AI agent without CRM integration is like a salesperson who never takes notes. Conversations happen, but nothing gets recorded, no follow-ups are scheduled, and insights are lost.

This guide covers everything you need to know about connecting AI agents to your CRM.

Why CRM Integration Matters

Without CRM Integration

• Conversations happen in a silo
• Agents don't know customer history
• No record of AI interactions
• Manual data entry required
• Follow-ups fall through cracks

With CRM Integration

• Full customer context in every conversation
• Automatic logging of all interactions
• AI updates records in real-time
• Triggers workflows and automations
• Unified customer view

Supported CRMs

Edesy integrates with major CRMs out of the box:

What AI Agents Can Do With CRM

Read Operations

AI can access:
├── Contact/Lead information
├── Company/Account data
├── Deal/Opportunity status
├── Past interactions and notes
├── Custom fields and properties
├── Related records and associations
└── Activity history

Write Operations

AI can update:
├── Create new contacts/leads
├── Update existing records
├── Log conversation notes
├── Create tasks and follow-ups
├── Update deal stages
├── Add tags and categories
└── Trigger automations

Integration Setup: Salesforce

Step 1: Connect Salesforce

1. Go to Integrations > CRM > Salesforce
2. Click "Connect Salesforce"
3. Log in with your Salesforce admin account
4. Grant permissions (read/write access)

Step 2: Map Fields

Configure which Salesforce fields your AI can access and update:

Contact Fields:
├── First Name → AI can read/write
├── Last Name → AI can read/write
├── Email → AI can read only
├── Phone → AI can read/write
├── Lead Source → AI can write
└── Custom fields → Configure individually

Step 3: Configure Actions

Set up what your AI should do:

On New Conversation:

• Look up contact by email
• Create contact if not found
• Pull relevant history

During Conversation:

• Access past interactions
• Check deal status
• View custom fields

After Conversation:

• Log conversation transcript
• Update contact record
• Create follow-up task
• Trigger workflow

Example: Lead Qualification Flow

1. User starts chat
2. AI asks qualifying questions (BANT)
3. AI creates/updates Lead in Salesforce
4. AI sets Lead Score based on responses
5. If qualified: Creates Opportunity, assigns to rep
6. If not qualified: Adds to nurture campaign
7. Logs full conversation as Activity

Integration Setup: HubSpot

Step 1: Connect HubSpot

1. Go to Integrations > CRM > HubSpot
2. Click "Connect HubSpot"
3. Select your HubSpot account
4. Approve requested permissions

Step 2: Configure Objects

Choose which HubSpot objects AI can access:

• Contacts: Customer records
• Companies: Organization data
• Deals: Sales opportunities
• Tickets: Support cases

Step 3: Set Up Actions

Contact Actions:

- Create Contact
- Update Contact Properties
- Add to List
- Enroll in Sequence
- Log Activity

Deal Actions:

- Create Deal
- Update Deal Stage
- Associate with Contact
- Set Deal Amount
- Log Notes

Ticket Actions:

- Create Ticket
- Update Ticket Status
- Assign to Owner
- Add Comments

HubSpot Workflow Triggers

Your AI can trigger HubSpot workflows:

Trigger: AI qualifies lead as "Hot"
Workflow:
├── Send internal notification
├── Create task for sales rep
├── Add to "Hot Leads" list
└── Enroll in fast-track sequence

Integration Setup: Custom CRM

If you use a custom or less common CRM, connect via REST API.

Step 1: Get API Credentials

From your CRM, obtain:

• API endpoint URL
• API key or OAuth credentials
• Documentation for endpoints

Step 2: Configure in Edesy

1. Go to Integrations > Custom API
2. Add new integration
3. Enter API details:

{
  "name": "My CRM",
  "baseUrl": "https://api.mycrm.com/v1",
  "auth": {
    "type": "bearer",
    "token": "{{CRM_API_KEY}}"
  }
}

Step 3: Define Actions

Create custom actions for each CRM operation:

Example: Create Contact

{
  "name": "Create Contact",
  "method": "POST",
  "endpoint": "/contacts",
  "body": {
    "firstName": "{{first_name}}",
    "lastName": "{{last_name}}",
    "email": "{{email}}",
    "phone": "{{phone}}",
    "source": "AI Agent"
  }
}

Best Practices

1. Start with Read-Only

Begin with read access only. Let AI pull customer data without writing. Once confident, enable write access.

2. Log Everything

Every AI conversation should create a record in your CRM. Future context depends on it.

3. Use Standard Fields

Map AI data to standard CRM fields when possible. Custom fields can create complexity.

4. Handle Duplicates

Configure duplicate detection rules. AI creating duplicate records is a common issue.

5. Set Up Error Handling

What happens when CRM API fails? Your AI should:

• Continue the conversation
• Queue the update for retry
• Alert your team

Real-World Examples

Example 1: Sales Qualification

Customer: "Hi, I'm interested in your enterprise plan"

AI Actions:
1. Looks up customer by email → Finds existing contact
2. Sees they're in "Prospect" stage
3. Notes: Last contacted 30 days ago
4. Previous interest: Professional plan

AI Response: "Welcome back, Sarah! I see you were
looking at our Professional plan last month. You
mentioned your team was growing - has that influenced
your interest in Enterprise?"

[Conversation continues with qualification questions]

AI Actions after conversation:
1. Updates Deal stage to "Qualified"
2. Logs conversation transcript
3. Creates task: "Schedule demo - Enterprise interest"
4. Triggers notification to account owner

Example 2: Support with Context

Customer: "My integration isn't working"

AI Actions:
1. Looks up customer → Enterprise account, $50K ARR
2. Checks recent tickets → 2 tickets last month
3. Checks product → Using API integration

AI Response: "I'm sorry to hear about the integration
issue, Michael. I see you're on our Enterprise plan
using the REST API. Is this related to the
authentication issue we resolved last week, or
something new?"

[Troubleshooting conversation]

AI Actions after resolution:
1. Creates ticket with transcript
2. Tags as "API - Authentication"
3. Updates customer health score
4. Logs resolution time for SLA tracking

Measuring Success

Track these metrics after CRM integration:

Troubleshooting

"AI can't find customer records"

• Check email matching logic
• Verify API permissions
• Review field mappings

"Duplicate records being created"

• Enable duplicate detection
• Match on multiple fields (email + phone)
• Use CRM's merge tools for cleanup

"Updates aren't syncing"

• Check API rate limits
• Verify write permissions
• Review error logs

"Wrong data being pulled"

• Check field mappings
• Verify object relationships
• Test with specific record IDs

Getting Started

Ready to connect your CRM?

1. Log in to Edesy
2. Go to Integrations > CRM
3. Select your CRM
4. Follow the connection wizard
5. Configure field mappings
6. Test with sample data
7. Go live

Need help? Contact our integration team.

Related Articles

• How to Build a Customer Support AI Agent
• Getting Started with AI Agent Builder
• AI Agent vs Chatbot: What's the Difference?