Building a No‑Code Drawing Tool with Coze AI Bot Platform: A Step‑by‑Step Guide

Introduction

Hello, I am yma16. This article shares the complete no‑code workflow for building a drawing tool with the Coze platform.

What is Coze?

Coze is a next‑generation, all‑in‑one AI Bot development platform offering four major features: rich plugin integration, a simple knowledge‑base for data management, persistent conversational memory, and a robust workflow engine.

Target audience includes data analysts, PPT writers, copywriters, UI designers, and anyone who prefers not to code.

From 0 to 1: Building the Drawing Bot

Enter the Coze workspace at https://www.coze.cn/space and create a new bot named "Drawing Tool".

Bot Details

Provide additional keywords in the tool description to automatically generate relevant icons.

Orchestration Area

The orchestration canvas consists of three sections: natural‑language logic (pseudo‑code), dependent functions (large model, plugins, I/O samples), and a preview/debug area.

Logic Definition

人物设定：ui设计师、画家
功能：根据用户关键词使用插件进行作画
工作流程：一对一交流

Dependency Configuration

Set the large‑model context length to 3 turns.

Add the "ByteArtist" drawing plugin via the "+" button in the top‑right of the canvas.

After adding the plugin, configure the reply card and its parameters.

Testing

Prompt the bot to "draw a tech‑style image with a dragon" and verify the generated result.

Adding a Workflow

Use the visual connector to define data flow between nodes.

Knowledge Base

Import custom data and define fields to store information for later retrieval.

Greeting and Voice

Set an initial greeting message for new users and choose a gentle "Peach" voice style.

Publishing

Publish the bot and view the live chat at https://www.doubao.com/chat/6056599270914 .

Analysis Area

The analysis view provides user‑state awareness, showing daily or weekly statistics of new and active users.

Conclusion

Coze is a low‑code platform that lets non‑technical users configure logic to invoke large models, effectively treating the model as a reusable component.

End

Thank you for reading; feedback is welcome.