Welcome to Lattice AIF, an AI framework for building intelligent applications. ## Lattice-AIF: New AI framework to create Agents Lattice-AIF is an agentic framework and platform designed to build, deploy, and manage intelligent AI agents with minimum effort. LatticeAI Engine provides a unified abstraction layer over diverse Large Language Models (LLMs) and external tools, empowering you to create robust, actionable AI solutions with unprecedented flexibility. The client-server architecture is inspired from Docker client and Docker engine, helps to scale and mange the Agents. ## HLD architecture diagram ![alt text](docs/images/arch.png) ## ✨ Why Lattice-AIF? In a world of fragmented AI capabilities, LatticeAI brings order and efficiency: * **LLM Agnostic:** Seamlessly integrate and switch between any LLM – from cloud-based powerhouses like Google Gemini and OpenAI to local, privacy-focused models like Ollama. * **Robust Tool-Calling:** Extend LLM capabilities beyond text generation. Connect your agents to any API, database, or internal system to perform real-world actions. * **Agent Orchestration:** Build complex, multi-step, multi-LLM agent workflows with ease, and manage their entire lifecycle. * **Lattice Engine:** Inspired by Docker Engine, the arch is designed to benefit from a continuously running, daemonized "Lattice Engine" that provides production-grade reliability, scalability, and resource management for your agents. * **Developer-Friendly:** Convert your existing Automation and Applications into AI tools without much workload along with MCP Integration. * **Memory Management:** Have a inbuilt memory management layer for the Agents. * **logging & Token Management:** Have a inbuilt logging and token management layer for the Agents. ## 📦 Package Structure Lattice-AIF is organized into three core packages to ensure modularity and scalability: - [**lattice-engine**](file:///home/pharsha/lattice-aif/lattice-engine/README.md): The core orchestration layer, which manages multiple connections to llm servers both cloud and local and also manage the tool/MCP servers. It acts as a interface between the client and the tools/llms. - [**lattice-client**](file:///home/pharsha/lattice-aif/lattice-client/README.md): The CLI layer for the user to interact with the engine. This package also includes a small streamlit UI application for demo. A standard UI web/desktop application is available as a seperate project([lattice-ui](https://github.com/trellisAI/lattice-ui)). - [**lattice-server**](file:///home/pharsha/lattice-aif/lattice-server/README.md): A lightweight utility for registering and exposing tools for agents. This provides a simple interface to expose tools to the engine using decorators to wrap existing python functions and classes. It provides alternative to MCP servers. ## 📖 Documentation Our comprehensive documentation covers everything from installation and configuration to advanced agent development and deployment strategies. **[Explore the Docs](https://trellisai.github.io/lattice-aif/docs.html)** ## 🤝 Contributing We welcome contributions from the community! Whether it's code, documentation, or ideas, your input is valuable. Please see our [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines. Join our community to connect with other developers, get support, and contribute to LatticeAI's growth: * **GitHub Issues:** Report bugs or suggest features. * **Discussions :** Ask questions and engage with the community. * **Discord (Coming Soon):** Real-time chat. ## 📄 License Lattice-AIF(TrellisAI Product) is released under the [MIT License](LICENSE).

Menu