xlsm-llm

Excel VBA functions for interacting with local and cloud-based LLMs, enabling text processing, translation, summarization, and code generation directly in Excel.
These functions allow you to send prompts to LLMs and retrieve AI-generated responses directly in your spreadsheet, supporting both local models and external APIs like OpenAI, Gemini, and Upstage.

Features

xlsm-llm provides powerful Excel VBA functions that seamlessly integrate with local and external LLMs, enabling text processing, translation, summarization, and code generation directly within Excel.

✨ Interact with LLMs directly in Excel

• LLM: Send a prompt and receive AI-generated responses in your spreadsheet.
• LLM_SUMMARIZE: Automatically generate concise summaries of long texts.
• LLM_CODE: Generate code snippets in any programming language based on your requirements.
• LLM_LIST: Extract structured lists from AI responses for better data organization.
• LLM_EDIT: Improve text by fixing grammar, punctuation, and clarity.
• LLM_TRANSLATE: Translate text into different languages, with custom translation prompts available.
• LLM_REVIEW_TRANSLATION: Evaluate and refine translations by comparing them to the original text.

⚡ Optimized for Ease of Use

• Works with both local LLMs and cloud APIs (e.g., OpenAI, Gemini, Upstage).
• API keys can be provided manually or set as environment variables for seamless authentication.
• Functions handle newline formatting automatically, ensuring clean Excel outputs.

✅ Reliable and Flexible

• Supports custom prompts for greater control over AI-generated outputs.
• Includes error handling for stable and predictable function execution.
• Compatible with various LLM models, including GPT-based models, Upstage Solar, and locally hosted models.

With xlsm-llm, you can harness AI capabilities without leaving Excel, enhancing productivity and automation in text processing and coding tasks.

Installation

1. Clone the repository:

   git clone https://github.com/ychoi-kr/xlsm-llm.git

2. Import the VBA modules into your Excel workbook:
   • Open your Excel workbook.
   • Press Alt + F11 to open the Visual Basic for Applications (VBA) editor.
   • In the VBA editor, press Ctrl + R if the Project Explorer is not already visible. Locate your workbook's project (e.g., VBAProject (YourWorkbookName.xlsm)).
   • Right-click on your workbook's project in the Project Explorer and select Import File….
   • Navigate to the src folder in the cloned repository, select the module file (for example, LLM_Functions.bas), and click Open.
   • The module should now appear in your Project Explorer.
3. Add the required Microsoft XML reference:
   • In the VBA editor, go to Tools > References.
   • Check Microsoft XML, v6.0 from the list and click OK.

Setting Environment Variables

Before using the functions, ensure that you have obtained an API key from your desired service provider (e.g., OpenAI, Gemini, Upstage, etc.). These keys allow you to authenticate with external LLM APIs. You can store these keys as environment variables so that the functions automatically use them when the API key parameter is omitted.

For example, if you are using the OpenAI API, set the environment variable OPENAI_API_KEY. Similarly, for Gemini or Upstage, you can set GEMINI_API_KEY or UPSTAGE_API_KEY, respectively.

• On Windows:
  • Open Control Panel > System > Advanced system settings > Environment Variables.
  • Under "User variables" (or "System variables" for all users), click New… and set the variable name (e.g., OPENAI_API_KEY) and the value as your API key.
  • Alternatively, use the command prompt:

    setx OPENAI_API_KEY your_api_key_here

• On macOS/Linux:
  • Edit your shell profile (e.g., ~/.bash_profile or ~/.zshrc) and add a line such as:

    export OPENAI_API_KEY=your_api_key_here

  • Then reload your profile by running source ~/.bash_profile (or the appropriate command for your shell).

This method works for any API key required by the functions; just replace OPENAI_API_KEY with the appropriate variable name for the service you are using (e.g., GEMINI_API_KEY, UPSTAGE_API_KEY).

Using a Local LLM

You can use xlsm-llm with a local LLM instead of a cloud-based API (e.g., OpenAI, Gemini, Upstage). To do this, you need to set up and run a local LLM server.

The easiest way to run a local LLM is with LM Studio:

1. Download and install LM Studio from lmstudio.ai.
2. Download a model – LM Studio suggests recommended models upon first launch, or you can browse and install a model of your choice.
3. Start the local server in the Local Server tab.
4. Set the base_url in Excel functions to match the server’s address (e.g., http://localhost:1234/v1/).
5. Specify the model explicitly when calling the function, as local LLMs require a model name.

For example:

=LLM("Write a haiku", "recursion in programming", 0.7, 100, "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

💡 No API key is required for most local LLMs.

For detailed setup instructions, refer to the LM Studio documentation.

Usage

LLM

Sends a prompt to an LLM (Large Language Model) and returns the AI-generated response. Useful for text generation, explanations, brainstorming, and creative writing.

Example:

=LLM("Write a short story about AI.")

➡ Generates a short AI-written story.

To add extra context, use the value parameter:

=LLM("Explain recursion", "using simple examples.")

For local LLMs, specify base_url and model:

=LLM("Describe quantum computing", , , , "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

📖 For detailed usage, visit:
🔗 LLM Wiki Page

LLM_SUMMARIZE

Condenses a given text into a shorter, more concise version. By default, it produces a one-line summary, but you can customize the format.

Example:

=LLM_SUMMARIZE(A1)

➡ Summarizes the text in A1 into a single sentence.

To change the summary style, use a custom prompt:

=LLM_SUMMARIZE(A1, "Summarize in bullet points:")

For local LLMs, specify base_url and model:

=LLM_SUMMARIZE(A1, "Summarize in simple terms.", , , "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

📖 For detailed usage, visit:
🔗 LLM_SUMMARIZE Wiki Page

LLM_CODE

Generates code snippets based on a given description and programming language. This function allows you to quickly create code directly within Excel.

Example:

=LLM_CODE("Calculate the factorial of a number", "Python")

➡ Generates a Python function to calculate the factorial of a number.

To generate JavaScript code:

=LLM_CODE("Create a basic countdown timer", "JavaScript")

For local LLMs, specify base_url and model:

=LLM_CODE("Sort a list using quicksort", "Python", "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

📖 For detailed usage, visit:
🔗 LLM_CODE Wiki Page

LLM_LIST

Generates a structured list based on a given prompt. The output is formatted as an array, making it easy to process in Excel.

Example:

=LLM_LIST("List three common fruits.")

➡ Returns: {"Apple", "Banana", "Cherry"}

To use a local LLM, specify base_url and model:

=LLM_LIST("List five AI models", "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

📖 For detailed usage, visit:
🔗 LLM_LIST Wiki Page

LLM_EDIT

Refines a given text by improving grammar, punctuation, clarity, and fluency. You can provide a custom prompt to adjust the editing style.

Example:

=LLM_EDIT(A1)

➡ Corrects grammar and improves readability of the text in A1.

If you need a specific editing style, use a custom prompt:

=LLM_EDIT(A1, "Make this text more formal.")

For local LLMs, specify base_url and model:

=LLM_EDIT(A1, "Rewrite in a professional tone.", , , "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

📖 For detailed usage, visit:
🔗 LLM_EDIT Wiki Page

LLM_TRANSLATE

Translates text into a specified language. You can provide a custom translation prompt for more control over the output.

Example:

=LLM_TRANSLATE(A1, "es")

➡ Translates the text in A1 into Spanish.

If a custom prompt is provided, targetLang and sourceLang are ignored:

=LLM_TRANSLATE(A1, , , "Translate this text using formal language.")

For local LLMs, specify base_url and model:

=LLM_TRANSLATE(A1, "ja", , , , , "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

📖 For detailed usage, visit:
🔗 LLM_TRANSLATE Wiki Page

LLM_REVIEW_TRANSLATION

Evaluates a translation by comparing it to the original text, providing feedback on accuracy, grammar, fluency, and overall quality. You can specify a focus area (e.g., "fluency", "technical terminology") for targeted review.

Example:

=LLM_REVIEW_TRANSLATION(A1, B1, "fluency")

➡ Reviews the translation in B1 with a focus on fluency, comparing it to the original text in A1.

Works with both local and cloud-based LLMs. For local models, specify base_url and model:

=LLM_REVIEW_TRANSLATION(A1, B1, , , , "qwen2.5-14b-instruct", "http://localhost:1234/v1/")

📖 For detailed usage, visit:
🔗 LLM_REVIEW_TRANSLATION Wiki Page

License

This project is licensed under the MIT License.