Privacy Policy — Apexsions AI

1. Executive Overview

Apexsions AI Inc. ("Apexsions", "we", "our", or "us") provides an autonomous agentic runtime and simulation platform powered by upstream foundation models, primarily Anthropic PBC ("Anthropic Claude"). We are strictly committed to customer data confidentiality, zero-data model training guarantees, and enterprise compliance under GDPR, CCPA, and SOC 2 Type II guidelines.

2. Zero Customer Data Training Guarantee

Key Commitment: Neither Apexsions nor our upstream AI infrastructure providers (including Anthropic) use customer prompts, agent instructions, world states, or player conversational data to train, retrain, or improve foundational AI models.

All API requests sent to our inference endpoints are processed transiently in volatile memory and logged only for latency, diagnostic error tracing, and billing token counting, unless explicitly retained by the customer under our Persistent Memory Vault feature.

3. Categories of Information We Collect

Account & Billing Identifiers: Name, work email address, company organization, and payment verification records.
API Telemetry & Operational Data: Token usage metrics, execution latency (TTFT), error codes, and high-level tool execution status.
Persistent World State Embeddings: When customers opt-in to our 200k context vector storage, vector embeddings of game chronicles are stored in customer-isolated, AES-256 encrypted PostgreSQL/pgvector clusters.

4. Security & Encryption Standards

Our infrastructure enforces defense-in-depth protocols:

In-Transit Encryption: All incoming and outbound traffic is encrypted via TLS 1.3 with strict modern cipher suites and HSTS enforcement.
At-Rest Encryption: All database volumes, vector indexes, and system snapshots are encrypted utilizing FIPS 140-2 validated AES-256 keys.
Access Controls: Role-based access control (RBAC), multi-factor authentication (MFA), and zero-trust VPC architecture for all internal engineering operations.

5. Sub-processors and Third-Party Disclosures

We partner with vetted, enterprise-grade cloud providers for core platform delivery:

Anthropic PBC: Foundation AI model inference provider (Claude 5.5 Model Family: Claude Sonnet 5.5, Haiku 5.5, Opus 5.5). Operating under strict Zero Data Retention / Commercial BAA.
Cloudflare Inc.: Global Edge CDN, DDoS mitigation, and SSL termination.
Amazon Web Services (AWS) & Google Cloud: Secure isolated cloud compute and relational database hosting.

6. Contact Data Protection Officer

For data deletion requests, GDPR Data Subject Access Requests (DSAR), or security audits, contact our security office directly: