AI Data Usage Policy

Effective Date: March 2, 2026
Last Updated: March 2, 2026
Version: 1.0

1. Introduction

This AI Data Usage Policy explains how Arunkumar Chaubey, an individual resident of India, doing business as Porcia, uses artificial intelligence (AI) in our Service and how we handle data processed by AI systems.

Porcia uses AI to power core features including email analysis, contract review, vendor intelligence, and automation. This policy provides transparency about our AI practices.

Contact Us:

• General: hello@porcia.org
• Privacy: privacy@porcia.org
• AI Questions: hello@porcia.org

2. AI Models We Use

2.1 AI Provider

Porcia uses Microsoft Azure OpenAI Service (powered by Azure AI Foundry) as our primary AI provider.

Why Azure OpenAI:

• Enterprise-grade security and compliance
• No training on customer data
• Data residency controls
• Transparent data handling policies

2.2 AI Models

Porcia uses AI models provided by Microsoft Azure OpenAI Service through Azure AI Foundry to power various features.

Model Selection: We select appropriate models based on task requirements, balancing accuracy, speed, and cost.

3. How We Use AI

3.1 Email Intelligence

What AI does:

• Analyzes emails to extract relevant information
• Identifies vendor names, pricing, and contract terms
• Identifies renewal dates and payment information
• Detects contract changes and important updates

Data sent to AI:

• Email subject line
• Email body text
• Email sender and recipient
• Email attachments (when relevant)

Human oversight:

• AI classifications are reviewed by you in the dashboard
• You can correct misclassifications
• Corrections help improve accuracy (but don't train the model)

3.2 Contract Analysis

What AI does:

• Extracts key terms (pricing, duration, renewal, termination)
• Identifies unfavorable clauses
• Summarizes contract highlights
• Compares terms to market benchmarks

Data sent to AI:

• Contract document content (PDF, DOCX)
• Vendor name and context
• Your instructions or questions

Human oversight:

• You must review all AI-extracted terms
• AI may miss clauses or misinterpret language
• Always verify against the original contract

3.3 Negotiation Strategy

What AI does:

• Analyzes your current contract and vendor relationship
• Generates negotiation talking points
• Drafts negotiation emails
• Suggests alternative terms and pricing

Data sent to AI:

• Current contract terms
• Vendor information
• Your negotiation goals
• Historical pricing data (anonymized benchmarks)

Human oversight:

• All AI-generated emails are drafts only
• You must explicitly review and approve before sending
• Porcia never sends emails on your behalf without approval
• You can edit, reject, or regenerate AI drafts

3.4 Vendor Intelligence

What AI does:

• Analyzes public vendor information
• Generates vendor summaries and insights
• Identifies similar vendors and alternatives
• Processes aggregated, anonymized pricing data

Data sent to AI:

• Public vendor information (websites, pricing pages)
• Aggregated, anonymized customer data
• Market research and industry reports

Human oversight:

• Vendor intelligence is informational only
• Always verify information independently
• AI may have outdated or incorrect information

4. Critical AI Commitments

4.1 No Training on Your Data

Your data is NOT used to train AI models.

• Azure OpenAI does not use customer inputs or outputs for model training
• Your emails, contracts, and data remain private
• AI models are pre-trained by Microsoft/OpenAI
• We do not fine-tune models on customer data

4.2 Human-in-the-Loop

AI never takes action without your approval.

• All AI-generated emails are presented as drafts
• You must explicitly click "Send" to send any email
• Porcia has no authority to commit you to any terms
• All decisions remain under your control

4.3 No Autonomous Actions

Porcia is not an autonomous agent.

• We do not send emails on your behalf
• We do not sign contracts or make commitments
• We do not make purchases or payments
• We provide recommendations only — you make all decisions

4.4 Data Minimization

We send only necessary data to AI.

• We strip unnecessary personal information before AI processing
• We use context windows efficiently to minimize data transfer
• We don't send entire email threads when only recent messages are relevant

5. AI Data Handling

5.1 Data Sent to AI

When you use AI features, the following data may be sent to Azure OpenAI:

• Email content and metadata
• Contract documents
• Vendor information
• Your instructions and prompts
• Historical context (previous emails, contracts)

5.2 Data Retention by AI Provider

Azure OpenAI:

• Does NOT retain customer prompts or outputs for model training
• May retain data for up to 30 days for abuse monitoring
• Data is encrypted in transit and at rest
• Data residency in Azure regions

Porcia:

• Logs AI prompts and outputs for 30 days for debugging
• After 30 days, logs are automatically deleted
• Logs are encrypted and access-controlled

5.3 Data Not Sent to AI

We do NOT send:

• Passwords or authentication credentials
• Payment card information
• Unnecessary personal information
• Data from other customers

6. AI Accuracy and Limitations

6.1 AI Is Not Perfect

AI may:

• Misclassify emails or extract incorrect data
• Miss important contract clauses
• Generate negotiation strategies that don't work
• Provide outdated or inaccurate vendor information
• "Hallucinate" (generate plausible but false information)

6.2 Your Responsibility

You must:

• Review all AI outputs before acting on them
• Verify extracted data against original sources
• Use AI recommendations as guidance, not gospel
• Apply your own judgment and expertise
• Not rely solely on AI for important decisions

6.3 Not Professional Advice

AI outputs are NOT:

• Legal advice (consult a lawyer for legal matters)
• Financial advice (consult a financial advisor)
• Guaranteed to be accurate or complete
• A substitute for professional review

7. AI Security

7.1 Encryption

• All data sent to AI is encrypted in transit (TLS 1.2+)
• AI provider encrypts data at rest (AES-256)
• API keys and credentials are encrypted and access-controlled

7.2 Access Controls

• Only authorized Porcia systems can access AI service
• No direct human access to AI API
• All AI requests logged and monitored

7.3 Abuse Prevention

• Rate limiting prevents excessive AI usage
• Content filtering prevents harmful prompts
• Monitoring detects unusual patterns

8. AI Transparency

8.1 AI Disclosure

We clearly indicate when content is AI-generated:

• AI-generated emails are labeled "AI Draft"
• AI-extracted data is marked with an AI icon
• AI recommendations include disclaimers

8.2 Explainability

Where possible, we explain:

• Why AI made a particular classification
• What data AI used to generate recommendations
• Confidence levels for AI outputs

8.3 User Control

You can:

• Choose whether to use AI features
• Regenerate AI outputs if unsatisfied
• Provide feedback on AI accuracy
• Opt out of specific AI features (contact support)

9. AI Ethics

9.1 Fairness

• We do not use AI to discriminate based on protected characteristics
• We monitor AI outputs for bias
• We use diverse training data (via Azure OpenAI)

9.2 Accountability

• Porcia is responsible for AI outputs, not the AI provider
• We investigate and address AI errors
• We provide support for AI-related issues

9.3 Privacy

• We minimize data sent to AI
• We do not use AI to profile individuals
• We respect user privacy in AI processing

10. Aggregated Data and AI

10.1 Benchmarking

We may use aggregated, anonymized data to:

• Improve vendor intelligence
• Provide pricing benchmarks
• Enhance AI recommendations

Anonymization process:

• Remove all identifying information (company names, individual names)
• Aggregate data across many customers
• Ensure data cannot be re-identified

Opt-out: Contact privacy@porcia.org to opt out of contributing to aggregated benchmarks.

10.2 No Individual Profiling

We do NOT use AI to:

• Profile individual users
• Make automated decisions about individuals
• Discriminate or target based on personal characteristics

11. Third-Party AI Services

11.1 Current AI Providers

11.2 Future AI Providers

If we add new AI providers, we will:

• Update this policy with 30 days' notice
• Ensure equivalent data protection commitments
• Provide opt-out options where feasible

12. AI for Minors

Porcia is a B2B service not directed at individuals under 16. We do not knowingly use AI to process data of children.

13. AI and GDPR

13.1 Legal Basis

We process data with AI based on:

• Contract Performance: Providing the Service you subscribed to
• Legitimate Interest: Improving the Service and providing insights
• Consent: Where required for optional AI features

13.2 Data Subject Rights

You can exercise GDPR rights regarding AI processing:

• Right of Access: Request AI prompts and outputs about you
• Right to Erasure: Request deletion of AI logs
• Right to Object: Object to AI processing (may limit Service functionality)
• Right to Human Review: Request human review of AI decisions

Contact privacy@porcia.org to exercise rights.

13.3 Automated Decision-Making

Porcia does NOT make solely automated decisions with legal or significant effects. All AI outputs require human review and approval.

14. Changes to AI Practices

We may update our AI practices as technology evolves. Material changes will be communicated via:

• Email notification (30 days' advance notice)
• In-app notification
• Update to this policy

15. AI Feedback

We welcome feedback on AI features:

• Accuracy Issues: Report incorrect AI outputs via in-app feedback
• Feature Requests: Suggest AI improvements at hello@porcia.org
• Concerns: Raise AI ethics or privacy concerns at privacy@porcia.org

Your feedback helps us improve AI quality and safety.

16. Contact Us

Questions about AI and data usage?

General AI Questions: hello@porcia.org
Privacy Concerns: privacy@porcia.org
Security Issues: security@porcia.org

Arunkumar Chaubey
Doing business as Porcia
C/13 Mangalmurti Society, Ghatkopar West
Mumbai, Maharashtra, India
Phone: +91 8097907763

17. Additional Resources

• Privacy Policy
• Terms of Service
• Security
• Data Processing Addendum

18. AI Provider Policies

For more information about our AI providers:

• Azure OpenAI: Azure OpenAI Data Privacy
• Microsoft AI Principles: Microsoft Responsible AI

Version History:

• v1.0 (March 2, 2026) - Initial release