How to Build Conditional AI Chatbot Responses Based on CRM Properties
Learn how to configure your AI chatbot to only respond to specific customers based on CRM properties like product type, subscription tier, or custom fields.
Not every customer should get the same AI chatbot treatment. Your VIP clients, enterprise customers, or users of specific products often need different handling than your general customer base.
Conditional AI chatbot routing — filtering responses based on CRM properties. Photo by Zulfugar Karimov on Unsplash.
This guide shows you how to make your AI chatbot check CRM properties before responding, so it only engages with the right customers. (In this example, we use HubSpot and Make.)
Why Conditional Responses Matter
Imagine you support multiple products, but your AI chatbot only knows about Product A. If it responds to Product B customers with irrelevant information, you create confusion and frustration.
Conditional logic lets you:
Route product-specific questions to specialized AI knowledge bases
Prevent AI from responding to customers who need human-only support
Adjust AI behavior based on subscription tier or account value
Filter by customer status, location, or any CRM property
The Basic Pattern
Every conditional AI response follows this flow:
Message arrives from customer
Look up customer in CRM
Check relevant property values
If match: AI responds
If no match: AI stays silent (or routes to human)
Real-World Example: POS Software Filtering
Let's say you support restaurants using different point-of-sale systems. Your AI knows Restaurant365, but customers using Toast or Square need different answers.
Here's how to filter:
Step 1: Structure Your CRM Property
In HubSpot, create a company property:
Property name: POS Software
Type: Dropdown select
Options: Restaurant365, Toast, Square POS, Other
Step 2: Build the Lookup in Make
When a message arrives:
Module: HubSpot > Search for Contact
Search by: Phone number = message.from
Then get the associated company:
Module: HubSpot > Get a Company
Company ID: contact.associatedcompanyid
Filter condition:
company.pos_software does not equal "Restaurant365"
OR company.pos_software is empty
→ Either end silently or create a task for human follow-up
Step 4: Generate Contextual AI Response
Now your AI prompt can reference the confirmed POS:
System prompt:
"You are a support assistant for Restaurant365 users.
The customer uses Restaurant365 POS.
Answer their question: message.content"
Multiple Property Checks
You can stack conditions for more sophisticated routing:
Path A - AI handles:
- POS Software = Restaurant365
- AND Subscription Status = Active
- AND Support Tier = Standard or Premium
Path B - Human required:
- POS Software = Restaurant365
- AND Subscription Status = Active
- AND Support Tier = Enterprise
Path C - No response:
- All other combinations
Handling Missing or Incomplete Data
What if a customer's company record doesn't have a POS Software value set?
Option 1: Conservative approach (safer)
Default to no AI response
Create a task to fill in missing data
Let humans handle until data is complete
Option 2: Aggressive approach
Ask the customer directly: "Which POS system do you use?"
Update CRM with their response
Then provide AI assistance
Option 3: General fallback
Respond with product-agnostic help
Link to documentation
Offer to connect with a specialist
Using Dynamic Knowledge Bases
Take it further by switching AI knowledge sources based on properties:
If POS = Restaurant365:
Knowledge base = "r365-support-docs"
If POS = Toast:
Knowledge base = "toast-support-docs"
If POS = Square:
Knowledge base = "square-support-docs"
This lets one AI system handle multiple product lines intelligently.
Customer Tier Examples
Filter by subscription value:
If Annual Contract Value > $50,000:
→ No AI response, route to dedicated account manager
If Annual Contract Value < $50,000:
→ AI handles with standard knowledge base
Filter by support package:
If Support Plan = "Premium" AND Business Hours = True:
→ AI responds with premium context
If Support Plan = "Premium" AND Business Hours = False:
→ AI responds + notify on-call premium support rep
If Support Plan = "Basic":
→ AI responds from basic knowledge base only
Geographic or Regulatory Filtering
If Company Country = "Germany" AND Topic contains "data privacy":
→ Route to GDPR-trained specialist, no AI
If Company Country = "United States" AND State = "California":
→ Include CCPA disclosure in AI response
Combining with Intent Detection
For maximum sophistication:
Detect the message intent (question, complaint, purchase inquiry)
Look up CRM properties
Route based on both:
If Intent = "Billing question" AND Payment Status = "Overdue":
→ Route to collections team, no AI
If Intent = "Technical support" AND Product = "Enterprise Plan":
→ AI responds + notify technical account manager
Testing Your Conditional Logic
Create test scenarios:
Test case 1: Known customer, matching property
Expected: AI responds with correct context
Test case 2: Known customer, non-matching property
Expected: No AI response
Test case 3: Unknown customer (not in CRM)
Expected: Fallback behavior triggers
Test case 4: Customer with empty property value
Expected: Conservative handling (no AI or data gathering)
Performance and Latency
Adding CRM lookups adds approximately:
500-1500ms for contact search
300-800ms for company property retrieval
Total delay: ~1-2 seconds before AI generates response.
For most use cases, this is acceptable. For time-critical scenarios, consider caching frequently accessed customer data.
Monitoring and Refinement
Track these metrics:
Filter match rate: What percentage of messages pass your filters?
False negatives: Messages that should have gotten AI but didn't
False positives: Messages that got AI but shouldn't have
Property completeness: How often are required fields empty?
Refine your filters monthly based on this data.
Ready to Build Smarter AI Chatbot Filtering?
Setting up conditional AI chatbot routing takes the right automation architecture — and the details matter. If you want help designing the CRM lookup logic, router paths, and AI prompt structure for your specific setup, book a free discovery call with Connex. We'll review your current CRM setup and map out exactly how the filtering should work.
Learn how to build a Notion CRM for investor tracking in commercial real estate, capturing preferences and using AI-assisted deal matching with a simple four-table system.
Notion custom agents switch to paid credits May 4, 2026. Learn how to audit your usage, narrow triggers, and optimize agent workflows before the meter starts.
Learn how to build a Notion custom agent that delivers personalized daily briefings to your team — pulling calendar events, tasks, and meeting context automatically.
