--- title: "LilyGo-Esp-Claw" url: "https://maker.wiznet.io/bruno/projects/lilygo-esp-claw/" markdown_url: "https://maker.wiznet.io/bruno/projects/lilygo-esp-claw/md" type: "UCC: User Created Content" author: "LoshuaiB" author_url: "https://github.com/Xinyuan-LilyGO/LilyGo-Esp-Claw" editor: "WIZnet" editor_url: "https://maker.wiznet.io/" original_author: "LoshuaiB" original_url: "https://github.com/Xinyuan-LilyGO/LilyGo-Esp-Claw" published: "2026-11-04" language: "en" likes: 0 views: 4 comments: 0 source: "WIZnet Makers (https://maker.wiznet.io/)" --- # LilyGo-Esp-Claw > ESP-Claw is a Chat Coding AI agent framework for IoT devices. Original author: LoshuaiB (source: https://github.com/Xinyuan-LilyGO/LilyGo-Esp-Claw) ## Article ![Lilygo T-Connect Pro Industrial IoT Controller propose Ethernet, WiFi, Bluetooth, Lora, RS232, RS485 et Can Bus](https://images.openai.com/static-rsc-4/4pVWcQcpSYH2twZYVJFLgaNYruW_T8na2A5aXb3CPZzrAptHnyoZP8EinzZQ1M5gyTJJzGhHkNOrlhPOhowiI-0aAOX9nFJR7sbf62dvdw8Wd5e0hwxze9hYoQ7NmBWB_cs4dzeJdUV2Sjc_4dvBCOpkSTES46Moht_-KwamoMjCCjHpSNbP_BOXqIgyRrP5?purpose=fullsize) ## LilyGO ESP-Claw: Bringing AI Agents to Industrial IoT with W5500 Ethernet **Explore how the WIZnet W5500, ESP32-S3, and AI Agent technology come together to create new possibilities for intelligent industrial IoT devices.** ### Introduction What if an industrial IoT device could understand instructions in natural language instead of relying entirely on predefined commands? Imagine asking a device to check equipment status, retrieve sensor information, or execute an automation task without manually navigating complex menus. This is the concept behind **LilyGO ESP-Claw**, an AI Agent framework designed for Espressif-based IoT devices. Developed by **LILYGO**, the project brings conversational AI capabilities to embedded hardware, allowing users to define device behavior through natural language and execute supported tasks using software tools. One of its supported platforms is the **LILYGO T-Connect Pro**, an industrial IoT development board combining an ESP32-S3 microcontroller with multiple wired and wireless communication interfaces, including the **WIZnet W5500 Ethernet controller**. By bringing AI Agent capabilities together with industrial communication hardware, ESP-Claw offers an interesting starting point for exploring intelligent connected devices. ### What Is ESP-Claw? ESP-Claw is a conversational AI Agent framework that allows embedded devices to interact with users and perform tasks through a combination of language model services, local tools, and automation logic. Instead of programming every interaction as a fixed sequence, developers can create systems that interpret instructions, select available capabilities, and execute supported operations. The ESP32 runs the Agent software, manages device resources, and coordinates tool execution. Language-model inference can be performed through external services using OpenAI-compatible or Anthropic-compatible APIs. This provides a way to combine the flexibility of modern AI models with the physical interfaces available on embedded devices. ### Hardware Platform: LILYGO T-Connect Pro ![LILYGO T-Connect Pro - LILYGO Wiki](https://images.openai.com/static-rsc-4/5qj_zCshELyrOqbxxMBAl5ZCk14k9U71c0HZ5-B73g-rg8rKlGkIfc956HsezjbDo678bOmx5SLbO9yC03OUEHMppviw3TXGARG9K6Ao_ezZMp925NOCDSJGNLpMhCb954TvtSjrxGV47hOMaJrVC8WQ2ztJBK28c1YvrI-v9dGs9MnDhUpuGoMFU53qALU7?purpose=fullsize) The T-Connect Pro is an ESP32-S3-based industrial IoT development board equipped with multiple communication interfaces. Its main hardware features include: | Component | Specification | | --- | --- | | Microcontroller | ESP32-S3-R8 | | Memory | 16 MB Flash, 8 MB PSRAM | | Ethernet | WIZnet W5500 via SPI | | Wireless | 2.4 GHz Wi-Fi, Bluetooth LE | | LoRa | SX1262 | | Industrial Interfaces | CAN, RS485, RS232 | | Display | ST7796 TFT with capacitive touch | | Output | 10 A relay | | Power | 12–24 V DC input | The combination of Ethernet, serial communication, LoRa, and local control hardware makes this platform particularly interesting for industrial automation and connected equipment. In August 2026, the LilyGO ESP-Claw project added T-Connect Pro board support, including configuration for the W5500 Ethernet interface and other onboard peripherals. ### Key Features of ESP-Claw #### 1. Conversational AI Agent ESP-Claw allows users to interact with their IoT devices through natural language. The Agent can interpret requests, identify available software tools, and coordinate actions according to the device's configured capabilities. This approach can simplify how developers create user interfaces for complex embedded applications. #### 2. Web-Based Management Console ESP-Claw includes a browser-based console for device configuration and interaction. Users can access features such as: AI chat and session management Network configuration and device status LLM provider and API settings Capability and tool management Automation configuration Filesystem management The Web Console provides a convenient environment for testing and configuring Agent-based applications. #### 3. Automation and Lua Scripting ESP-Claw supports event routing, scheduled tasks, and Lua scripting. These capabilities allow developers to create custom automation behaviors and extend the functionality of the Agent. For example, a developer could build an application that reads equipment status, evaluates predefined conditions, and triggers a response through an available hardware interface. Actual device control functions must be implemented and configured for the target hardware. #### 4. Multiple Communication Interfaces The T-Connect Pro brings several industrial communication interfaces to the ESP-Claw ecosystem. These include: **W5500 Ethernet:** Wired network connectivity **CAN:** Communication with CAN-based controllers **RS485:** Interface for industrial serial equipment **RS232:** Connectivity with legacy serial devices **LoRa:** Long-range wireless communication This combination creates opportunities for AI-assisted gateways that communicate with different types of industrial devices. #### 5. Messaging Platform Integration ESP-Claw also supports integration with messaging services such as Telegram, WeChat, QQ, and Feishu. These interfaces provide additional ways to interact with the Agent without requiring users to access the device's local Web Console. ### The Role of WIZnet W5500 One of the important hardware features of the T-Connect Pro is its WIZnet W5500 Ethernet controller. The W5500 integrates a hardware TCP/IP engine and a 10/100 Ethernet interface, allowing microcontrollers to add wired network connectivity through SPI. For an AI-enabled IoT system, Ethernet can be particularly valuable when a device operates in a fixed industrial environment. Wired Ethernet offers several practical advantages: **Stable Connectivity** Industrial equipment frequently operates in environments where wireless signals may be affected by interference, physical obstacles, or network restrictions. Ethernet provides an alternative connection path for installations requiring wired communication. **Integration with Existing Networks** Many industrial facilities already use Ethernet infrastructure to connect controllers, monitoring systems, and gateways. Adding Ethernet connectivity can help IoT devices fit into these environments. **Connectivity for Cloud-Based AI Services** AI Agent applications often communicate with external APIs to obtain language-model responses or exchange data with other systems. Ethernet can provide a suitable network interface for this communication. In ESP-Claw, W5500 support is included in the T-Connect Pro hardware configuration. The published project demonstrates board-level support, while the complete AI communication workflow over W5500 should be validated on the target firmware before deployment. ### Architecture: Connecting AI and Industrial Devices A possible application architecture using ESP-Claw and T-Connect Pro can be organized as follows: **User Interaction** Web Chat / Messaging Applications ↓ **ESP-Claw Agent on ESP32-S3** Instruction interpretation, tool selection, and automation ↔ **External LLM API** Language understanding and reasoning ↓ **Device Capabilities** Lua scripts, event routing, and hardware interface drivers ↓ **Industrial Interfaces** CAN / RS485 / RS232 / LoRa / Relay The W5500 provides a potential wired Ethernet connection for the network-dependent portions of this system. This architecture illustrates how an embedded AI Agent could act as a bridge between user instructions, cloud-based AI services, and industrial equipment. ### Potential Applications #### AI-Assisted Industrial Gateway An AI-enabled gateway could allow operators to request equipment information using natural language. For example: *“Check the current operating status of the connected controller.”* The Agent could translate the request into a supported application command, retrieve the appropriate data, and return a readable response. #### Intelligent Equipment Monitoring By combining communication interfaces with automation scripts, developers could create monitoring systems that summarize equipment status and report abnormal conditions. #### Natural-Language Industrial Control With appropriate hardware drivers and authorization controls, the platform could support natural-language interaction with industrial equipment. Commands affecting physical equipment would require additional safety checks, access controls, and deterministic execution logic. #### AI-Enabled Sensor Gateways The combination of LoRa, Ethernet, and serial interfaces could support gateways that collect sensor data from multiple communication networks and make that information accessible through conversational interfaces. These examples represent potential applications of the platform rather than fully validated end-to-end demonstrations supplied by the project. ### Getting Started The LilyGO ESP-Claw repository provides firmware resources and instructions for ESP32-S3-based devices. A typical setup process involves: Prepare a supported LILYGO T-Connect Pro board. Obtain the appropriate firmware image from the project repository. Flash the firmware using Espressif's Flash Download Tool, Web Flasher, or esptool. Restart the device. Connect to the device's configuration access point. Open the Web Console and configure Wi-Fi and the desired LLM provider. Test the Agent using Web Chat. Explore the available capabilities and hardware interfaces. For Ethernet-based applications, developers should verify that the selected firmware initializes the W5500 and that the required network services operate over the Ethernet interface. **Security note:** The ESP-Claw documentation describes its Web Console as a development-oriented interface that does not use HTTPS by default. It should be used in a trusted environment and must not be exposed directly to the public Internet. Additional authentication, secure provisioning, and device-control safeguards are necessary for production deployments. ### Project Resources **Project:** LilyGO ESP-Claw **Developer:** LILYGO **Supported Board:** T-Connect Pro **Ethernet Controller:** WIZnet W5500 **GitHub Repository:** **T-Connect Pro Hardware Documentation:** **T-Connect Pro Product Page:** ### Conclusion LilyGO ESP-Claw offers an interesting example of how conversational AI can be introduced into embedded IoT systems. By combining an ESP32-S3-based AI Agent framework with industrial communication interfaces, the project provides a foundation for exploring intelligent gateways, equipment monitoring, and natural-language device interaction. The inclusion of the WIZnet W5500 on the T-Connect Pro highlights the potential role of wired Ethernet in AI-enabled industrial IoT applications. As AI Agents continue to evolve, projects such as ESP-Claw demonstrate how embedded networking and physical device interfaces can become part of more intelligent, connected systems. --- Source: https://maker.wiznet.io/bruno/projects/lilygo-esp-claw/