Arcanflows — AI Automation Platform for Business

Overview

Form integration connects your AI agents directly to forms, enabling intelligent processing of submissions, conversational form experiences, and automated data handling.

Integration Modes

1. Post-Submit Processing

Agent processes form data after submission:

┌──────────┐    ┌──────────┐    ┌──────────┐
│   Form   │───▶│ AI Agent │───▶│  Action  │
│ Submitted│    │ Process  │    │ (Email,  │
└──────────┘    └──────────┘    │  DB, etc)│
                               └──────────┘

2. Conversational Form

Agent guides user through form completion:

┌──────────┐    ┌──────────┐    ┌──────────┐
│   User   │◀──▶│ AI Agent │───▶│   Form   │
│   Chat   │    │ Collect  │    │   Data   │
└──────────┘    └──────────┘    └──────────┘

3. Hybrid Mode

Traditional form with AI assistance:

┌──────────────────────────────────────────┐
│               Form UI                     │
├──────────────────────────────────────────┤
│  [Name: _________]                        │
│  [Email: ________]                        │
│  [Question: _____]     ┌──────────────┐  │
│                        │  AI Chat     │  │
│  [Submit]              │  Assistant   │  │
│                        └──────────────┘  │
└──────────────────────────────────────────┘

Configuration

Basic Form-Agent Connection

json
{
  "form": {
    "id": "contact_form",
    "agent_integration": {
      "enabled": true,
      "agent_id": "support_agent",
      "mode": "post_submit",
      "process_fields": ["message", "question"]
    }
  }
}

Form Settings

Setting	Type	Description
`agent_id`	string	Agent to connect
`mode`	string	"post_submit", "conversational", "hybrid"
`process_fields`	array	Fields to send to agent
`include_all_fields`	boolean	Send all form data
`response_action`	string	What to do with response

Post-Submit Processing

Configuration

json
{
  "agent_integration": {
    "mode": "post_submit",
    "agent_id": "form_processor",
    "message_template": "Process this form submission:\n\nName: {{name}}\nEmail: {{email}}\nMessage: {{message}}",
    "response_action": "store",
    "store_field": "ai_analysis"
  }
}

Use Cases

Lead Qualification

json
{
  "agent_integration": {
    "mode": "post_submit",
    "agent_id": "lead_qualifier",
    "message_template": "Qualify this lead:\n\nCompany: {{company}}\nRole: {{job_title}}\nBudget: {{budget}}\nTimeline: {{timeline}}\nNeeds: {{requirements}}",
    "response_format": {
      "type": "json",
      "schema": {
        "score": "number",
        "qualification": "string",
        "next_action": "string"
      }
    },
    "post_process": {
      "update_submission": {
        "lead_score": "{{response.score}}",
        "qualification": "{{response.qualification}}"
      },
      "trigger_workflow": {
        "workflow_id": "lead_routing",
        "condition": "{{response.score}} >= 70"
      }
    }
  }
}

Support Ticket Classification

json
{
  "agent_integration": {
    "mode": "post_submit",
    "agent_id": "ticket_classifier",
    "message_template": "Classify this support request:\n\nSubject: {{subject}}\nDescription: {{description}}\nUrgency (user-selected): {{urgency}}",
    "response_format": {
      "type": "json",
      "schema": {
        "category": "string",
        "priority": "string",
        "department": "string",
        "suggested_response": "string"
      }
    },
    "post_process": {
      "create_ticket": {
        "system": "zendesk",
        "mapping": {
          "subject": "{{subject}}",
          "description": "{{description}}",
          "category": "{{response.category}}",
          "priority": "{{response.priority}}",
          "assignee_group": "{{response.department}}"
        }
      },
      "send_auto_response": {
        "condition": "{{response.suggested_response}} !== null",
        "email": "{{email}}",
        "template": "auto_response",
        "variables": {
          "suggested_answer": "{{response.suggested_response}}"
        }
      }
    }
  }
}

Feedback Analysis

json
{
  "agent_integration": {
    "mode": "post_submit",
    "agent_id": "feedback_analyzer",
    "message_template": "Analyze this customer feedback:\n\nRating: {{rating}}/5\nFeedback: {{feedback}}\nProduct: {{product}}",
    "response_format": {
      "type": "json",
      "schema": {
        "sentiment": "string",
        "themes": "array",
        "actionable_insights": "array",
        "requires_followup": "boolean"
      }
    }
  }
}

Conversational Forms

Configuration

json
{
  "form": {
    "id": "intake_form",
    "mode": "conversational",
    "agent_id": "intake_agent",
    "conversation_config": {
      "greeting": "Hi! I'll help you get started. What's your name?",
      "fields_to_collect": [
        {
          "field": "name",
          "prompt": "Nice to meet you! What's your name?",
          "validation": "required"
        },
        {
          "field": "email",
          "prompt": "What's the best email to reach you at?",
          "validation": "email"
        },
        {
          "field": "company",
          "prompt": "What company are you with?",
          "validation": "optional"
        },
        {
          "field": "needs",
          "prompt": "Tell me about what you're looking for today.",
          "validation": "min_length:20"
        }
      ],
      "completion_message": "Thanks {{name}}! I've got everything I need. Someone from our team will reach out to {{email}} shortly."
    }
  }
}

Agent System Prompt for Conversational Forms

markdown
You are a friendly intake specialist collecting information from potential customers.

Your goal is to collect the following information naturally through conversation:
- Name
- Email
- Company (optional)
- Their needs/requirements

Guidelines:
1. Be conversational and friendly, not robotic
2. Ask one question at a time
3. Acknowledge their responses before asking the next question
4. If they provide multiple pieces of information at once, acknowledge all of them
5. Validate email format before accepting
6. For "needs", encourage them to be specific

When you have all required information, summarize what you've collected and confirm.

Use the collect_field tool to store each piece of information:
- collect_field(name, value) - stores the field value

Example conversation:
User: "Hi, I'm interested in your product"
You: "Great to hear! I'd love to help you learn more. First, what's your name?"
User: "I'm John from Acme Corp"
You: [collect_field("name", "John"), collect_field("company", "Acme Corp")]
     "Nice to meet you, John! And what's the best email to reach you at Acme Corp?"

Embed a conversational form on your website:

html
<script src="https://cdn.arcanflows.com/form-widget.js"></script>
<script>
  Arcanflows.conversationalForm({
    formId: 'intake_form',
    container: '#form-container',
    theme: 'light',
    position: 'inline' // or 'floating'
  });
</script>

Hybrid Mode

Combine traditional form with AI assistance:

Configuration

json
{
  "form": {
    "id": "application_form",
    "mode": "hybrid",
    "agent_config": {
      "agent_id": "form_assistant",
      "position": "sidebar",
      "trigger": "help_button",
      "capabilities": [
        "answer_questions",
        "explain_fields",
        "suggest_values",
        "validate_input"
      ]
    }
  }
}

AI Assistance Features

Field Help

json
{
  "fields": {
    "annual_revenue": {
      "type": "number",
      "label": "Annual Revenue",
      "ai_help": {
        "enabled": true,
        "prompt": "Explain what we mean by annual revenue and why we need it",
        "examples": true
      }
    }
  }
}

Smart Suggestions

json
{
  "fields": {
    "job_title": {
      "type": "text",
      "label": "Job Title",
      "ai_suggest": {
        "enabled": true,
        "based_on": ["company_type", "department"],
        "prompt": "Suggest appropriate job titles for someone in {{department}} at a {{company_type}}"
      }
    }
  }
}

Validation Assistance

json
{
  "fields": {
    "description": {
      "type": "textarea",
      "label": "Project Description",
      "ai_validate": {
        "enabled": true,
        "rules": [
          {
            "check": "completeness",
            "prompt": "Check if this description includes: goals, timeline, budget, and success criteria"
          },
          {
            "check": "clarity",
            "prompt": "Evaluate if this description is clear and specific enough"
          }
        ],
        "feedback_mode": "inline"
      }
    }
  }
}

Response Actions

Store Response

Save AI response with submission:

json
{
  "response_action": {
    "type": "store",
    "field": "ai_analysis",
    "format": "json"
  }
}

Trigger Workflow

Start a workflow with the response:

json
{
  "response_action": {
    "type": "workflow",
    "workflow_id": "process_application",
    "input": {
      "form_data": "{{submission}}",
      "ai_response": "{{response}}"
    }
  }
}

Send Notification

Notify based on response:

json
{
  "response_action": {
    "type": "notification",
    "conditions": [
      {
        "if": "{{response.priority}} === 'urgent'",
        "channel": "slack",
        "message": "Urgent form submission from {{name}}: {{response.summary}}"
      },
      {
        "if": "{{response.score}} >= 80",
        "channel": "email",
        "to": "[email protected]",
        "subject": "High-value lead: {{name}}"
      }
    ]
  }
}

Custom Webhook

Send to external service:

json
{
  "response_action": {
    "type": "webhook",
    "url": "https://your-service.com/api/form-processed",
    "method": "POST",
    "body": {
      "submission_id": "{{submission.id}}",
      "form_data": "{{submission.data}}",
      "ai_analysis": "{{response}}"
    }
  }
}

Embedding Options

Inline Form with Agent

html
<div id="arcanflows-form"></div>
<script src="https://cdn.arcanflows.com/forms.js"></script>
<script>
  Arcanflows.renderForm({
    formId: 'contact_form',
    container: '#arcanflows-form',
    agentAssistant: {
      enabled: true,
      position: 'right-sidebar',
      collapsed: false
    }
  });
</script>

Floating Chat Form

html
<script src="https://cdn.arcanflows.com/forms.js"></script>
<script>
  Arcanflows.conversationalForm({
    formId: 'intake_form',
    mode: 'floating',
    position: 'bottom-right',
    triggerButton: {
      text: 'Get Started',
      icon: 'chat'
    }
  });
</script>

API Integration

Submit Form to Agent

bash
curl -X POST "https://api.arcanflows.com/api/v1/forms/{form_id}/submit" \
  -H "X-API-Key: your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "data": {
      "name": "John Doe",
      "email": "[email protected]",
      "message": "I need help with..."
    },
    "process_with_agent": true
  }'

Response

json
{
  "submission_id": "sub_abc123",
  "status": "processed",
  "data": {
    "name": "John Doe",
    "email": "[email protected]",
    "message": "I need help with..."
  },
  "agent_response": {
    "category": "support",
    "priority": "medium",
    "suggested_response": "Thank you for reaching out..."
  }
}

Best Practices

1. Clear Agent Instructions

Define exactly what the agent should do with form data:

json
{
  "message_template": "You are processing a support request form. Analyze the submission and provide:\n1. Category (billing/technical/general)\n2. Priority (low/medium/high/urgent)\n3. Suggested response\n\nForm data:\n{{json submission}}"
}

2. Handle Validation

Let the agent help with validation:

json
{
  "pre_submit_validation": {
    "agent_id": "validator",
    "check_fields": ["description", "requirements"],
    "validation_prompt": "Check if these fields are complete and clear enough for us to help"
  }
}

3. Graceful Fallbacks

Handle agent failures gracefully:

json
{
  "error_handling": {
    "on_agent_error": "submit_without_processing",
    "fallback_classification": {
      "category": "general",
      "priority": "medium"
    },
    "notify_admin": true
  }
}

4. Privacy Considerations

Be mindful of what data is sent to agents:

json
{
  "privacy": {
    "exclude_fields": ["ssn", "credit_card"],
    "mask_fields": ["phone"],
    "anonymize": false
  }
}

5. Test Thoroughly

Test the integration with various inputs:

Valid submissions
Edge cases
Invalid data
Large text inputs
Special characters

Overview

Integration Modes

1. Post-Submit Processing

2. Conversational Form

3. Hybrid Mode

Configuration

Basic Form-Agent Connection

Form Settings

Post-Submit Processing

Configuration

Use Cases

Lead Qualification

Support Ticket Classification

Feedback Analysis

Conversational Forms

Configuration

Agent System Prompt for Conversational Forms

Conversational Form Widget

Hybrid Mode

Configuration

AI Assistance Features

Field Help

Smart Suggestions

Validation Assistance

Response Actions

Store Response

Trigger Workflow

Send Notification

Custom Webhook

Embedding Options

Inline Form with Agent

Floating Chat Form

API Integration

Submit Form to Agent

Response

Best Practices

1. Clear Agent Instructions

2. Handle Validation

3. Graceful Fallbacks

4. Privacy Considerations

5. Test Thoroughly