Guspora^™

1. Overview

This AI Transparency Notice explains how the Guspora platform uses artificial intelligence. Guspora uses AI to reduce inspection documentation time and improve report quality. AI is a tool that assists licensed professionals. It does not replace professional judgment, perform inspections, or make compliance determinations.

1A. Default Operating Configuration

Guspora AI Agents are engineered with specific safety guardrails ("Default Configuration"):

• Action risk classification per agent action category
• Autonomy ceilings per agent
• Human approval requirements for high-risk actions
• Communication consent enforcement for TCPA and CAN-SPAM compliance
• Audit logging through the Agent Action Ledger
• Tenant-scoped data isolation

Guspora accuracy claims, service descriptions, and warranties for AI Agents apply only when Agents operate in Default Configuration.

1B. Customer-Modified Configuration

Customers enter Modified Configuration when they lower approval thresholds, expand autonomy ceilings, modify or disable consent enforcement, expand data scope, or direct AI Agents outside intended use cases. Outcomes caused by Modified Configuration are the customer's responsibility.

2. AI Systems in Use

3. Human Review Requirement

The technician is responsible for:

• Verifying the accuracy of AI-generated findings
• Confirming code citations are correct and applicable
• Ensuring manufacturer-specific terminology is accurate
• Adding, removing, or modifying any content as needed
• Approving the final report for delivery

AI-generated content is a draft. The licensed technician is the author of record.

4. AI Content Marking

Reports containing AI-generated content include a disclosure statement indicating that AI assistance was used in report preparation and that the report was reviewed and approved by a licensed technician.

5. Data Protection

• Inspection data sent to AI providers is transmitted over encrypted connections (TLS 1.2+)
• AI providers process data solely for the requested purpose
• AI providers are contractually prohibited from retaining data or using it for training
• No personally identifiable information is included in AI prompts unless necessary for the specific function

6. Jurisdiction-Specific Guardrails

AI-generated content is configured to respect jurisdiction-specific requirements:

• Kentucky home inspection reports (per KRS 198B.738) do not include code compliance statements or purchase recommendations
• Reports are configured for the applicable code edition and local amendments
• Manufacturer-specific terminology databases ensure accurate equipment references

Service providers are ultimately responsible for verifying compliance with their jurisdiction.

7. Limitations

• AI Agents are probabilistic systems and may produce inaccurate or unexpected outputs even in Default Configuration
• AI may not be aware of the most recent code amendments or local ordinances
• AI-generated code citations should be verified against the applicable code edition
• AI cannot assess conditions it cannot observe (hidden defects, environmental factors)
• AI transcription accuracy depends on audio quality, background noise, and accents
• AI is not a substitute for hands-on inspection by a qualified professional
• Customers must verify outputs before relying on them for legal, financial, regulatory, compliance, safety-critical, or professional decisions

8. Agent Action Ledger

The Agent Action Ledger records AI Agent activity, including action category, risk class, actor or approving user, tenant scope, target record, approval status, timestamp, and related metadata. The Ledger is the authoritative record in disputes about what an Agent drafted, recommended, initiated, routed, sent, or executed.

9. Customer Responsibilities

• Communications drafted or sent by AI Agents on a customer's behalf are customer communications
• Billing, payment, or other financial actions initiated by AI Agents are customer actions
• Compliance assessments and AHJ communications are advisory customer communications
• Customers must maintain lawful TCPA consent, CAN-SPAM compliance, data privacy compliance, and professional licensing where applicable

10. Your Choices

• Technicians may opt out of AI report generation and write reports manually
• All AI-generated content is fully editable before delivery
• Contact support if you require that no AI processing be applied to your inspection data