Report Abuse

Basic Information

DeliteAI is an on-device AI platform designed for building agentic workflows and embedding AI-native experiences into mobile applications. The repository provides the core components, SDKs, and examples developers need to integrate AI agents into Android and iOS apps while keeping inference and orchestration on-device for privacy and performance. It includes a C++ core runtime with Python bindings, a Python interface for orchestrating agents and tool calls, example agents and mobile sample apps, and documentation and community guidance to get started. The project aims to enable secure, low-latency AI features on resource-constrained devices and to streamline development of agent-driven mobile functionality through unified APIs and cross-platform abstractions.

Links

App Details

Features
Unified and simplified APIs for integrating AI agents into Android and iOS applications. A Python interface for orchestrating complex agentic workflows, including tool calling, memory management, and local LLM usage. Cross-platform portability and an abstracted development layer to support fragmented device landscapes. Optimizations for resource-constrained environments with efficient CPU and memory usage. Privacy-first, on-device processing with hardware-accelerated model execution. Extensibility through custom Python operators and flexible runtime support, including ONNX and ExecuTorch. Included artifacts and guidance for building the core C++ SDK and Python bindings, plus SDKs and sample apps for mobile testing and exploration.
Use Cases
DeliteAI helps developers deliver secure, privacy-aware, and high-performance AI features on mobile devices by enabling inference and orchestration to run locally. It reduces latency and dependence on remote services, supports constrained hardware through runtime optimizations, and simplifies cross-platform development via unified APIs and SDKs. The repository provides a practical starting point with a core runtime, Python bindings, mobile SDKs and sample apps, and an agents directory to explore ready-made examples. Documentation and community channels support adoption and contribution. Extensibility options let teams add custom operators or target different runtimes to fit specific device capabilities or application needs.

Please fill the required fields*