Show HN: Conduit – local-first desktop LLM client, byok (github.com)

Conduit, a new local-first desktop client for large language models (LLMs), has been announced, allowing users to communicate with various AI providers while ensuring their data remains on their personal devices. Built on Tauri 2 with a Rust backend and React frontend, Conduit enables users to integrate with eleven different LLM providers by securely storing API keys in the operating system's keychain. The application emphasizes user privacy by foregoing accounts and telemetry, encrypting all data with AES-256-GCM, and using a strict content security policy to prevent unauthorized data transmission. This innovative tool signifies a shift towards enhancing privacy and security in AI interactions, particularly for users concerned about data leaks or surveillance. By employing a local SQLite database for storing conversations and generated artifacts, Conduit offers encryption at rest, ensuring sensitive information is protected. Additionally, features like consent-gated tool usage and localized processing minimize the risk of data exposure while maximizing user control over their AI experiences. The project's open-source nature invites contributions, fostering collaboration within the AI/ML community.