The buzz surrounding Large Language Models (LLMs) has been impossible to ignore, with ChatGPT leading the charge in capturing public imagination. While the potential applications of these AI marvels seem boundless, one area generating particular excitement is their use in data analysis and spreadsheet manipulation. However, it's crucial to approach this integration with a measured perspective.

LLMs aren't quite ready for full deployment with spreadsheets just yet. When tasked with complex analysis, they can produce errors that might go unnoticed by the untrained eye. That said, dismissing their utility entirely would be a mistake. LLMs can be incredibly useful in your spreadsheet work in various ways that don't rely on their analytical capabilities. Let's explore some of the best ways to leverage LLMs like ChatGPT in conjunction with your spreadsheet tasks.

1. ChatGPT: Your AI-Powered Excel Tutor

The world of spreadsheets is vast and complex. Excel boasts over 500 functions, while Google Sheets offers more than 400. Whether you're a novice, intermediate, or advanced user, chances are there are some tasks in Excel that you don't know how to do. Typically, you'd turn to a YouTube tutorial or a quick Google search to find the answer.

This is where ChatGPT shines as your personal spreadsheet sensei. As your coach, master, teacher, or tutor - call it what you will - ChatGPT is there to educate you on the way of the cell. Here's how it can help:

• Function Finder: Need a specific, obscure function? ChatGPT can quickly provide you with the right one, complete with usage examples.

• Formula Debugging: Struggling with a complex formula? ChatGPT can help you find those missing brackets in your lengthy nested IF statement.

• Code Optimisation: Have a clunky, nested IF statement that needs streamlining? ChatGPT can suggest more elegant solutions, explaining the rationale behind its choices.

The beauty of this approach is that ChatGPT doesn't need direct access to your data. You can simply copy and paste formulas or describe your problem, allowing you to learn and grow as you progress from Excel white belt to black belt.

2. Effortless Data Analysis with ChatGPT

With its connection to OneDrive and Google Drive, ChatGPT can now easily import CSV and XLS files. Alternatively, you can upload files directly into the chat interface. This capability, combined with its advanced data analysis tools and code interpreter, allows ChatGPT to interact with and manipulate your data in powerful ways.

Data processing and visualisation become remarkably simple with ChatGPT. For instance, you could combine five separate spreadsheets into one master sheet - a task that might typically take hours of manual work, especially when dealing with historical exchange rate data across multiple currencies. With ChatGPT, this can be accomplished in minutes.

However, it's important to note that ChatGPT isn't infallible. It might provide a text interpretation of the data that contains inaccuracies. For example, I've encountered instances where graphs were incorrectly described in the analysis on almost every data analysis task I've performed with ChatGPT. Always double-check its interpretations against your raw data.

3. Try Using Google Gemini with Google Sheets

Google's answer to ChatGPT, Gemini, is another powerful tool in the LLM arsenal. Powered by Gemini Advanced, it's highly capable of manipulating data files such as CSVs and Google Sheets. It excels at complex analysis, particularly when dealing with data spread across multiple files.

In a recent project, I uploaded sales data from a restaurant to Gemini. It not only analysed the data but also created a customer segmentation strategy for a marketing automation plan. It even suggested implementing a Recency, Frequency, Monetary (RFM) value analysis, complete with recommended tactics for each customer segment.

4. Use Copilot in Microsoft Excel

Microsoft's version of ChatGPT, Copilot, has been integrated directly into Excel, offering a seamless experience for Office users. Copilot can answer questions about your data, suggest formulas to help you use the spreadsheet more effectively, and even add content directly into your spreadsheet.

For example, Copilot can add a formula into a specific cell or create an entire column of data in your table. In one instance, I had a column of date of birth data but needed the actual age as of the current day for each record. I asked Copilot how to do this, and it not only provided the formula but also added the entire column in a matter of seconds.

However, like all AI tools, Copilot isn't perfect. I've encountered situations where it implemented formatting on requested data but didn't apply the right conditions, despite confidently replying "Done!" Additionally, large datasets can prove challenging - sheets with over 2 million cells can result in the AI assistant simply being unable to work on the document.

5. Explore Rows.com for API-First Spreadsheets

Rows.com offers a reimagined approach to spreadsheets, focusing on API-first functionality. This platform connects to major services such as Notion, HubSpot, GA4, Instagram, Google Ads, Search Console, Google My Business, LinkedIn Ads, Looker Studio, Salesforce, Mailchimp, and many more.

In addition to the usual spreadsheet functions, Rows.com provides custom-built functions for specific integrations. For example:

• =SEARCH_DEALS_HUBSPOT: List and get details about HubSpot deals

• =SEARCH_CONTACTS_HUBSPOT: List and get details about contacts in your HubSpot

• =SEARCH_COMPANIES_HUBSPOT: List and get details about companies in your HubSpot

• =VIEW_SMS_TWILIO: Get the status of an SMS sent using Twilio

What sets Rows.com apart is its integration with OpenAI, offering custom functions that enable AI prompts for specific business use cases. Some of these functions include:

• ASK_OPENAI: Answer questions or complete tasks such as translation & copywriting

• REWRITE_OPENAI: Use OpenAI's GPT to rewrite text according to user-defined instructions

• KEYWORDS_OPENAI: Generate advertising keywords from a product description

• FIND_FACT_OPENAI: Discover facts about a country, person, company, or the world

• EXTRACT_OPENAI: Extract an element from text

• META_DESCRIPTION_OPENAI: Generate meta descriptions for a page using a list of keywords

• SENTIMENT_ANALYSIS_OPENAI: Run sentiment analysis on text

• SUMMARIZE_OPENAI: Summarize text into different types

• KEYWORDS_MISSPELLING_OPENAI: Generate common misspellings from a list of keywords

• AD_CONTENT_OPENAI: Generate different types of advertising content from a product description

• PRODUCT_TITLE_OPENAI: Generate a product title based on a description

• PRODUCT_DESCRIPTION_OPENAI: Generate a product description from a set of features

• CLASSIFY_OPENAI: Classify text into tags, topics, or sentiment

• TRANSLATE_OPENAI: Translate text between languages

• CREATE_LIST_OPENAI: Generate lists of items based on a prompt

• APPLY_TASK_OPENAI: Clean up or apply logic rules to data

Rows.com also features a built-in AI analyst that can provide insights and answer questions about your data within the tool. While this feature is somewhat limited, the deep integration with OpenAI and dozens of other services makes Rows.com particularly interesting for tasks such as content creation, sales development, sentiment analysis, translation, and more.

6. Supercharge Google Sheets with AI Add-Ons

Google Sheets add-ons offer another avenue for integrating LLM capabilities into your spreadsheet work. These add-ons can create entirely new spreadsheet functions connected to various LLMs.

For instance, Anthropic offers an official add-on that connects Claude (their AI model) to Google Sheets, creating a new function =CLAUDE(). This function allows you to interact with Claude directly from your spreadsheet cells, enabling tasks like text generation, question answering, and data analysis.

Similarly, there are add-ons available that connect to other AI models and services. For example, an add-on connecting the Perplexity AI search engine (which is powered by leading LLMs) creates a new function =PERPLEXITY(). This function can be used to search web pages, provide summaries of content, draft emails, categorise content, optimise web pages, and much more.

These add-ons essentially turn your spreadsheet into a powerful AI-augmented tool, allowing you to harness the capabilities of various LLMs without leaving your familiar spreadsheet environment.

Conclusion

While LLMs like ChatGPT aren't yet ready to completely take over your spreadsheet analysis, they offer a wealth of possibilities to enhance your work. From serving as your personal spreadsheet tutor to helping with data manipulation, from powering custom functions to enabling advanced integrations, LLMs are reshaping how we interact with and utilize spreadsheets.

As with any new technology, it's crucial to approach these tools with a balance of enthusiasm and caution. Always verify the outputs, especially for critical tasks, and be aware of the limitations of each tool. With this mindset, you can leverage the power of LLMs to significantly boost your productivity and capabilities in spreadsheet work.

Remember, the field of AI and LLMs is rapidly evolving. Stay curious, keep exploring new tools and integrations, and you'll be well-positioned to take advantage of the exciting developments that are sure to come