Terms of Service

01. Unified Terms of Agreement

Welcome to Prism AI (the “Platform”). By participating in the early preview sandbox, loading developer schemas, or evaluating orchestration instances, you accept and agree to undergo these binding Terms of Service. If you evaluate Prism AI representatively for an organization, you affirm that you have deep authority to bind them cleanly.

02. Credentials and BYOC Policy

Prism AI operates underneath a strict Bring Your Own Credentials (BYOC) architecture. The platform facilitates direct pipelines to models running in your personal or company cloud (such as Google Vertex AI, Anthropic, or OpenAI).

• Key Security responsibility: You are fully, singularly responsible for the cost, rotation, and safety of keys used to authenticate inside your browser application.
• Rate Limiting: Any rate constraints, tokens burned, or financial obligations resulting from executing sandboxed prompt commands fall entirely within your independent subscription tiers.

03. Permitted Sandbox Use & Safety

Since the operating platform relies heavily on compiling code blocks inside isolated microservice layers, you explicitly resolve not to utilize Prism AI to:

• Deploy malicious software, network scanning utilities, or outbound intrusion routines.
• Circumvent token caps, trigger model hallucinations intentionally seeking systemic backdoors, or attempt database injection routines.
• Host unauthorized command-and-control operations within our virtualized preview spaces.

04. Disclaimer of Guarantee

Prism AI is provided on an “AS IS” and “AS AVAILABLE” model. Since we are in Phase 1: Architecture & Planning of active releases, we issue zero certifications, guarantees, or durability claims on volatile variables or execution success rates.