AI Transparency & Data Usage Policy

Effective Date: September 6, 2026

Scope: Operational guidelines governing AI inference, model interactions, prompt processing, and data privacy across all ForgeCore multi-agent simulations.

2.1 Zero Model Training Guarantee

ForgeCore utilizes enterprise commercial API endpoints provided by OpenAI (gpt-5.6-terra, gpt-5.6-luna). Under our strict API integration architecture:

• No Public Training: Your proprietary input data, problem statements, strategic contexts, and generated Master Plans are NEVER used to train, retrain, or improve public AI models.
• Data Isolation: All prompt contexts submitted to OpenAI API endpoints are isolated to the specific API session and are governed by enterprise zero-data-retention standards.

2.2 Pre-Execution Content Retrieval (Jina Reader)

• When a Client provides a target URL within a task specification, our system invokes a pre-execution scraper (Jina Reader) to convert public HTML into clean Markdown.
• This process retrieves only publicly accessible web data. It does not bypass authentication, paywalls, or restricted access portals.

2.3 Human-in-the-Loop & Quality Assurance

ForgeCore operates as a Managed AI Studio:

• Pre-Handling: Expert human engineers review and refine task statements, inject required context, and map optimal domain-expert AI agent personas prior to simulation execution.
• Post-Handling: Expert human architects review, edit, and synthesize the generated AI Master Plan and code blueprints prior to final PDF export and client delivery.
• No Autonomous Action: AI agents operate purely within isolated simulation environments and cannot execute live actions, commit code, or alter production systems outside the simulation.

ForgeCore Platform is designed and built by Webtech Solutions.