法的情報
Data Handling & Security
最終更新: July 2026
Core Philosophy
- Insight-Aware, Not Data-Retentive: We prioritize extracting human understanding and organizational knowledge over the long-term retention of your raw information.
- Transient Processing: We interact with your data only when necessary to perform analysis and discard raw inputs immediately upon processing.
Definitions
Source Identifiers
The unique locations where you store your intellectual property or work-in-progress (e.g., Git repositories, LMS instances, document workspaces).
Source Data
The raw content contained within your connected platforms (e.g., source code, course materials, legal discovery documents) that Ninchi analyzes.
Derived Information
Proprietary analytics (e.g., Ninchi Scores, knowledge maps) created by our engine that do not contain raw Source Data.
How Ninchi Works
- Trigger: An activity occurs within a connected Source Identifier (e.g., a file update, a submission, or an access request).
- Analysis: The system securely accesses the relevant Source Data subset to generate context-aware questions or evaluations.
- Response: The user (e.g., developer, student, professional) provides an answer.
- Evaluation: Ninchi evaluates the response against the current context.
- Cleanup: Raw Source Data is discarded immediately after the analysis is complete.
What We Do NOT Store
What We DO Store (Derived Information Only)
Security & Compliance
SOC 2 (in progress): Ninchi is currently in the process of obtaining SOC 2 certification. Our compliance posture is continuously monitored with Vanta and Drata, and audit logs and compliance metadata are reviewed to verify our internal controls. We're happy to share our current security posture and certification progress with prospective enterprise customers under NDA.
Infrastructure
- Encryption in transit (TLS) and at rest (AES-256)
- Secure cloud infrastructure (AWS) with strict access controls
- Private networking for backend services
- Secure secret management and least-privilege service accounts
Enterprise Deployment: Dedicated deployment options, including dedicated VPCs and isolated tenant environments, can be arranged under an Enterprise agreement to satisfy strict data residency and egress requirements.
AI & Model Usage
- Zero-Training Pledge: We do not train external AI models (including foundational or third-party models) on any customer Source Data.
- Data Siloing: Your data is never shared across organizations. Analysis is strictly focused on patterns and understanding within your specific environment, not on proprietary logic transfer.
Proprietary Analytics
- Ninchi Score: A transparent, difficulty-weighted record of demonstrated understanding. It is Derived Information — a summary of verified interactions with your organization's work, not a measure of anyone's ability or intelligence, and not a copy of your raw input.
- Knowledge Maps: Visual representations of organizational understanding. These are generated from aggregated metadata and contain no raw Source Data.
Designed for Trust
Our architecture is built on a "minimal retention" model. By focusing on verified human understanding rather than raw data storage, we ensure that your intellectual property remains under your control while providing the actionable insights needed to scale your organization.
Contact / Support
For security inquiries, contact us at privacy@ninchi.ai.