Quick Start

This section will guide you through quickly completing the first runnable workflow of Agent for LabVIEW. Content is organized around the current two main modules:

  • LLM: Chat, context, tool calling

  • ImageGeneration: Text-to-image, image-to-image, image-text fusion

Pre-Start Checklist

Please confirm the following conditions are met first:

I. LLM Minimum Runnable Workflow

Objective: Get the Agent to chat normally and return results.

  1. Open example: Help -> Find Examples -> Directory Structure -> VIRobotics -> AI Agent

  2. Select the service provider with configured API Key

  3. Select basic chat example (xxbasic.vi), this manual uses Zhipu as example

  4. Enter a simple instruction (e.g., "Who are you")

  5. Click Run and observe the returned result

  6. Select streaming basic chat example (xxstream.vi), run, enter a simple instruction, click "send", observe the returned result

II. Comprehensive Example: AI Agent Full.vi

AI Agent Full.vi is the most important comprehensive example in the LLM section and should be run immediately after basic and streaming chat verification.

This example typically covers:

  • Streaming output

  • Tool calling (Tool Calling)

  • Multi-step task execution

Can be run as follows:

  1. Open AI Agent Full.vi and run

  2. Check tool calling useTools

  3. Select tools for this session, then enter natural language descriptions containing tool intent (e.g., "Help me parse this VI"). To select all tools, click "All". To uncheck all tools, click "None".

  4. Observe whether streaming output and tool calling execute as expected.

Note: Currently, ImageGeneration capabilities can be invoked by LLM as a tool, so the "understand task -> invoke image capability -> result output" loop can be verified in this comprehensive example. However, remember to activate and configure Doubao's related APIs and models.

Criteria for success:

  • Can receive model text response

  • No License errors

  • No API authentication errors

II. ImageGeneration Quick Start

It is recommended to verify this module in the order: text-to-image -> image-to-image -> image-text fusion.

2.0 Environment Requirements (Cannot run if not met)

Currently, ImageGeneration is primarily based on Doubao capabilities. Please complete the following first:

  1. Activate Doubao doubao-seedream model

  2. Complete Doubao API configuration in the Agent

  3. Install Yiqi Intelligence's AI Vision Toolkit for GPU, download: https://www.virobotics.net/product_AIVT_GPU

If the above conditions are not met, it is recommended to skip this section and prioritize completing LLM-related example verification.

2.1 Text-to-Image

Objective: Generate images through text.

  1. Open example: Help -> Find Examples -> Directory Structure -> VIRobotics -> AI Agent -> Doubao -> doubao_text_to_image_test.vi

  2. Enter image description in Prompt (subject, style, composition), set necessary parameters (size, quality, style)

  3. Run and check output image

Example prompts:

Generate an industrial laboratory scene image, containing oscilloscopes, robotic arms, and monitoring screens, realistic style.

2.2 Image-to-Image

Objective: Redrawing, variant, or style transfer based on existing images.

  1. Run VI: doubao_image_to_image.vi, load input image

  2. Enter editing description (e.g., "Put a red hat on the dog in the photo.")

  3. Run and compare input and output images

2.3 Image-Text Fusion

Objective: Complete targeted editing combining images and text instructions.

  1. Run VI: doubao_image_fusion.vi, load input image

  2. Add fusion text constraints

  3. Run generation

  4. Check content consistency and readability

III. LLM + ImageGeneration串联示例

Objective: Complete a minimum multi-module collaboration workflow.

Example task: "Generate an experimental workflow diagram and provide three usage instructions."

Recommended execution chain:

  1. LLM parses task and generates image prompts

  2. Call ImageGeneration to produce image

  3. Pass image results back to LLM to supplement explanatory text

  4. Output final results (image + instructions)

Common Issues and Recommendations

1) LLM can respond, but image generation fails

  • Check if Doubao API is activated

  • Check image interface quota and permissions

  • Shorten prompts and reduce complex constraints, then retry

2) basic runs successfully the first time, but no response the second time

  • Check if API key still has balance

3) basic.vi runs normally, but stream.vi returns nothing

  • If API is confirmed to be recharged, please contact Yiqi team at support@virobotics.net and tell us your LabVIEW version and computer system version.

Tips and Precautions

  • It is recommended to complete LLM verification first before entering ImageGeneration

  • It is recommended to use English paths to avoid potential compatibility issues with Chinese/spaces

  • When interface calls fail, prioritize checking: License status, API Key, network connectivity

  • For detailed module descriptions, refer to Module Overview

Technical Support

If you encounter issues during use, please first consult the relevant sections or contact technical support: