Privacy Policy

Effective Date: November 13, 2026 | Version: 2.2 (Enterprise Zero-Retention Architecture)

At Kernax, privacy is the engineering foundation of our ecosystem. Unlike conventional AI development tools that accumulate vast data lakes of user codebases for model training, Kernax is explicitly architected around a Zero-Retention philosophy. This Privacy Policy details how we protect your intellectual property, what minimal telemetry we collect, and how your data is processed.

1. Introduction

Kernax ("we", "us", "our") provides an advanced Artificial Intelligence Integrated Development Environment (IDE) and related cloud services (collectively, the "Services"). This Privacy Policy applies to all users ("you", "your") accessing our website, desktop applications, APIs, and associated services.

2. The Zero-Retention Architecture

We recognize that source code is the most critical asset of any modern enterprise. Our infrastructure functions strictly as a secure, stateless proxy layer:

• No Source Code Storage: Code snippets, workspace context, and file structures transmitted to our APIs are processed entirely in memory (RAM). Once the AI inference response is successfully streamed back to your local IDE, the payload is irrevocably purged from our servers.
• No Prompt Storage: Conversational instructions, queries, and terminal logs integrated into your prompts are not persisted to database systems.
• No AI Fine-Tuning: Kernax categorically does not utilize your codebase or conversational history to train, fine-tune, or reinforce proprietary or open-source language models.

3. Data We Collect

While we refuse to store proprietary workspace data, the operational and commercial realities of our platform require the minimal collection of Account and Billing information.

3.1 Account Information & Security Telemetry

To establish and secure your account, we collect your email address, securely hashed passwords (bcrypt), and authentication tokens. Additionally, to protect trial allocations and prevent automated abuse, sybil attacks, or unauthorized multi-account exploitation, our clients may generate and transmit cryptographic device-level anti-abuse tokens.

3.2 Billing & Usage Metadata

To process usage-based billing and maintain accurate wallet balances, we record non-content telemetry strictly limited to computational consumption:

• Input, output, and cached token volumes (e.g., 4,500 input tokens, 850 output tokens).
• The selected model identifier (e.g., kernax-auto, claude-sonnet-4.6, or other supported models).
• The corresponding usage deduction (in USD) applied to your Kernax wallet balance in accordance with Kernax’s active platform rates.

Note: Usage logs contain zero source code, file paths, or prompt text. Aggregated transaction and billing records are retained only as necessary for account reconciliation, fraud prevention, and statutory tax compliance.

4. How We Use Your Data

The minimal data acquired is utilized strictly for the following purposes:

• Service Provision: Authenticating your IDE client and connecting it securely to our computational grid.
• Financial Accounting: Deducting exact balances based on token utilization and generating corporate invoices.
• Fraud Prevention: Identifying automated botnets, credential stuffing, and Terms of Service violations.
• Platform Optimization: Monitoring holistic API latency and orchestrating load-balancing strategies across AI providers.

5. Third-Party AI Subprocessors & Inference Routing

Kernax securely routes inference workloads through vetted commercial API gateways and direct enterprise endpoints. All inference traffic is strictly governed by zero-retention commercial terms (including data_collection: deny enforcement), ensuring that neither Kernax nor our inference subprocessors persistently store, log, or train foundation models on your code:

• Fireworks AI, Inc.: High-throughput, low-latency commercial inference gateway for frontier open-weights models, operating under commercial zero-retention and zero-training policies.
• OpenRouter (OpenRouter Inc.): Secure commercial API routing gateway enforcing non-logging, ephemeral payload processing with strict data_collection: deny headers across upstream foundation providers.
• Anthropic, PBC: Enterprise API endpoints for Claude models, governed by dedicated commercial non-training and ephemeral inference protocols.
• OpenAI, L.L.C.: Enterprise API endpoints for GPT-series models, with data processing addendums prohibiting secondary model training and persistent prompt retention.
• DeepSeek AI: Dedicated inference endpoints for DeepSeek models under strict zero-persistence commercial privacy SLAs.

All payload transmissions between your IDE, the Kernax proxy gateway, and upstream subprocessors are encrypted in transit via Transport Layer Security (TLS 1.3).

6. IDE Telemetry & Analytics

The Kernax Desktop Application (based on Microsoft's OSS VS Code architecture) may collect anonymous crash reports and application performance indices (Telemetry). This telemetry strictly excludes source code and is used solely to identify software instability.

You may effortlessly opt-out of all non-essential IDE telemetry by modifying the setting telemetry.telemetryLevel to off within your workspace configuration.

7. Global Data Protection Rights (GDPR & CCPA)

As a global engineering platform, Kernax adheres to leading international data protection frameworks, including the European Union General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA / CPRA).

• Right to Erasure (Right to be Forgotten): You may terminate your account via your Dashboard at any moment. Invoking this right instantly triggers a cascading cryptographic deletion of your identity, balance, and association with historical metrics.
• Right to Portability: You may request an export of your account details and billing ledger in a machine-readable JSON format.
• DPA (Data Processing Agreement): For enterprise clients requiring a formal signing of a standard Data Processing Addendum (DPA), please contact our security team.

8. Security & Local Nodes

The Kernax Desktop client orchestrates a highly isolated background Intelligence Node operating physically on your machine. This Node is purposefully bound to the 127.0.0.1 (Localhost) network interface, rendering it mathematically invisible and inaccessible to malicious actors on global or local network perimeters.

All remote communications traversing the internet between your IDE and the Kernax Cloud are secured using Transport Layer Security (TLS 1.3) cryptographic protocols.

9. Contacting Security & Privacy

If you represent an enterprise compliance organization or have inquiries regarding our data processing, security architecture, or privacy practices, our team can be reached directly at:

Security & Privacy Inquiries: [email protected]
Legal & Compliance: [email protected]
Department: Kernax Security & Privacy