Kantos

Kantos AI Classification automatically analyzes incoming form submissions and routes them to the appropriate objects, assigns tags, and enriches data—all without manual intervention.

AI-Powered Lead Management

Stop manually sorting leads. AI Classification understands context, intent, and content to intelligently categorize every submission in milliseconds.

How It Works

When a form submission arrives via the Universal Script, the AI Classification engine:

Analyzes Content - Examines all form fields, values, and metadata
Identifies Intent - Determines if it's a sales inquiry, support request, partnership proposal, etc.
Routes Automatically - Sends the record to the appropriate object and assigns relevant tags
Enriches Data - Adds classification metadata for reporting and automation triggers

Enabling AI Classification

Navigate to Smart Intake Settings

In your Kantos dashboard, go to Settings → Smart Intake.

Enable AI Classification

Toggle on "AI Classification" in the processing options.

Configure Classification Rules

Set up categories, target objects, and routing rules based on classification results.

Test with Sample Submissions

Submit test forms to verify classifications are working as expected.

Classification Categories

Out of the box, AI Classification recognizes common business categories:

Category	Description	Common Indicators
`sales_inquiry`	Purchase interest or pricing questions	pricing, demo, buy, quote, purchase
`support_request`	Help with existing product/service	help, issue, problem, not working, error
`partnership`	Business partnership proposals	partner, collaborate, integration, reseller
`job_application`	Employment inquiries	resume, CV, position, hiring, job
`media_press`	Press and media inquiries	interview, article, press, media, journalist
`general_inquiry`	Catch-all for other inquiries	Default when no specific category matches

Custom Categories

Define your own classification categories to match your business needs:

{
  "custom_categories": [
    {
      "name": "wholesale_inquiry",
      "description": "Bulk purchase or wholesale account requests",
      "keywords": ["wholesale", "bulk", "distributor", "reseller", "volume"],
      "target_object": "obj_def_wholesale_leads"
    },
    {
      "name": "franchise_interest",
      "description": "Franchise ownership inquiries",
      "keywords": ["franchise", "territory", "location", "own a"],
      "target_object": "obj_def_franchise_leads"
    },
    {
      "name": "event_booking",
      "description": "Event or venue booking requests",
      "keywords": ["book", "event", "venue", "party", "reservation"],
      "target_object": "obj_def_event_requests"
    }
  ]
}

Category Training

The AI learns from your corrections. When you manually reclassify a submission, the system improves its future accuracy for similar submissions.

Routing Rules

Configure what happens when each category is detected:

{
  "routing_rules": [
    {
      "category": "sales_inquiry",
      "actions": {
        "target_object": "obj_def_sales_leads",
        "assign_tags": ["hot-lead", "needs-followup"],
        "notify_users": ["sales-team@company.com"],
        "trigger_automation": "auto_sales_welcome"
      }
    },
    {
      "category": "support_request",
      "actions": {
        "target_object": "obj_def_support_tickets",
        "assign_tags": ["unassigned"],
        "set_status": "new",
        "trigger_automation": "auto_support_acknowledgment"
      }
    },
    {
      "category": "partnership",
      "actions": {
        "target_object": "obj_def_partnerships",
        "assign_tags": ["review-needed"],
        "notify_users": ["partnerships@company.com"]
      }
    }
  ]
}

Classification Confidence

Every classification includes a confidence score (0-100). Use this to set thresholds:

{
  "confidence_settings": {
    "high_confidence_threshold": 85,
    "low_confidence_action": "flag_for_review",
    "fallback_category": "general_inquiry"
  }
}

High Confidence (85+) - Auto-route without review
Medium Confidence (50-84) - Route but flag for verification
Low Confidence (<50) - Send to review queue or fallback category

Classification Metadata

Every classified submission includes metadata you can use in automations and reports:

{
  "classification": {
    "category": "sales_inquiry",
    "confidence": 92,
    "secondary_categories": [
      {"category": "partnership", "confidence": 34}
    ],
    "detected_intent": "pricing_request",
    "suggested_tags": ["enterprise", "urgent"],
    "processed_at": "2024-01-15T10:30:00Z"
  }
}

Using Classification in Automations

Build automations that trigger based on classification results:

{
  "automation": {
    "name": "High-Value Lead Alert",
    "trigger": {
      "event": "record.created",
      "conditions": [
        {"field": "classification.category", "operator": "equals", "value": "sales_inquiry"},
        {"field": "classification.confidence", "operator": "greater_than", "value": 80},
        {"field": "data.company", "operator": "is_not_empty"}
      ]
    },
    "actions": [
      {"type": "send_email", "template": "high_value_lead_alert"},
      {"type": "create_task", "assignee": "sales_manager", "due": "+1 hour"}
    ]
  }
}

Multi-Object Routing

A single form can be classified into different objects based on content:

Configure Default Object

Set a default object in your Universal Script for unclassified submissions.

Set Up Category-Object Mapping

Map each classification category to its target object.

Enable Dynamic Routing

Turn on "Dynamic Object Routing" in Smart Intake settings.

Field Mapping

When routing to different objects, AI Classification automatically maps form fields to matching object fields. Fields without matches are stored in a "raw_data" JSON field.

Improving Classification Accuracy

1. Provide Context in Forms

Include fields that help the AI understand intent:

<form>
  <!-- Subject line provides clear context -->
  <select name="inquiry_type">
    <option value="">What can we help with?</option>
    <option value="pricing">Pricing & Plans</option>
    <option value="support">Technical Support</option>
    <option value="partnership">Partnership Inquiry</option>
    <option value="other">Other</option>
  </select>

  <!-- Message body for AI analysis -->
  <textarea name="message" placeholder="Tell us more..."></textarea>
</form>

2. Review and Correct Classifications

In your Kantos dashboard, review flagged submissions and correct misclassifications. The AI learns from these corrections.

3. Update Keywords Regularly

As your business evolves, update category keywords to reflect new products, services, or terminology your customers use.

Classification Reports

Track classification performance in your dashboard:

Category Distribution - See how submissions break down by category
Confidence Trends - Monitor average confidence over time
Correction Rate - Track how often manual corrections are needed
Routing Accuracy - Measure if leads reach the right destination

Privacy & Data Handling

AI Classification processes form data securely:

Data is analyzed in real-time and not stored for training by default
Classification happens within your Kantos tenant
PII is handled according to your data retention policies
You can disable AI Classification for specific forms if needed

Next Steps

Enable spam protection to filter out unwanted submissions
Create automations that trigger based on classification
Set up email templates for category-specific responses