Official Agreement & Data Manifesto

Terms & Conditions

Please read our authentic commitment to how your communications, OAuth tokens, and enterprise customer data are handled with zero guesswork.

Effective Date: August 31, 2026Version 2.4.0

“We treat your customer emails like physical sealed envelopes. We never read them for ads, never train AI models on your private messages, and enforce human approval whenever high-stakes actions are detected.”

— The Niyojitaa Engineering & Security Team

1. Mailbox Access & OAuth2 Token Governance

When you connect Google Workspace (Gmail) or Microsoft 365 Outlook to Niyojitaa EmailAgent, our platform interacts exclusively via official OAuth 2.0 application scopes (e.g., gmail.modify, Mail.ReadWrite).

Zero Password Retention: We never receive, prompt for, or store your raw account passwords.

AES-256-GCM Token Encryption: Access tokens and refresh tokens are encrypted at rest using dedicated AWS KMS keys.

Instant Revocation: You can disconnect and flush active tokens at any time with a single click in your Mailbox Settings.

2. Real-Time PII Masking & Zero AI Model Training

We believe your customer emails are strictly confidential. To prevent any leakage of sensitive information to foundational language models:

  • PII Scrubber: Social security numbers, credit card tokens, personal phone numbers, and physical addresses are masked or tokenized prior to context evaluation.
  • Zero Training Commitment: Your emails, thread histories, internal knowledge docs, and agent execution logs are never used to train, fine-tune, or improve public LLM models.
  • Ephemeral Processing: Ingestion pipelines process threads in memory and retain parsed metadata solely for audit logs and workbench review.

3. Deterministic Guardrails & Human-in-the-Loop

The agent acts under strict policy guardrails configured by your administrators. You agree and understand that:

High-Confidence Actions

Standard inquiries with high RAG relevance and zero risk triggers may be answered autonomously according to your policy thresholds.

Human Review Fallback

Contract modifications, billing refunds, escalations, or low confidence drafts are quarantined into the Approval Workbench for human sign-off.

4. Multi-Tenant Isolation & Audit Trails

All tenant accounts are strictly isolated at both the database and cache layers using cryptographic tenant IDs. Cross-tenant data leakage is strictly prohibited by our architectural firewall. Every email triage event, AI draft revision, policy evaluation, and manual human approval is written to an immutable audit trail.

5. Acceptable Use & Anti-Spam Policy

You agree not to utilize Niyojitaa EmailAgent for unsolicited bulk emailing (spam), automated credential harvesting, impersonation, phishing, deceptive marketing, or any purpose violating the CAN-SPAM Act, GDPR, or relevant international telecommunication laws.

6. Service Availability & Support SLA

We target a 99.9% uptime SLA across all webhook ingestion, background workers, and dashboard workbench systems. Enterprise tiers include dedicated support conduits and customized compliance agreements.

Acknowledged & Executed By:

hari7261

Lead Architect & Platform Engineering

✓ 100% Privacy VerifiedSOC2 • GDPR • Zero-Training
Questions regarding our terms or custom enterprise DPA agreements? Contact our Legal & Compliance Desk