This article is more than 1 year old
Microsoft's Garage band offers album of experimental Excel jazz
For those who think spreadsheet app isn't wild enough, try some of these potential future widgets
Microsoft is going to let organizations try out experimental Excel projects it's working on.
Redmond this week is releasing Excel Labs, an avenue to let spreadsheet users try out ideas the software giant is working on as part of Microsoft's larger Garage scheme, a long-running hackathon project Redmond runs around the world to encourage direct code development.
"With Excel Labs, in alignment with the Garage's mission, expect to find very early-stage ideas that we are thinking about and wanting to evaluate with our customers," said Petra Ronald, program manager II at Microsoft, adding that even though some ideas won't become part of Excel, the feedback from enterprises is important.
Organizations can install the Excel Labs add-in via the Office Store, though they need to ensure their systems meet the necessary requirements to try the first of two Excel projects that Microsoft is allowing access to.
The first is the Advanced Formula Environment (AFE), which is an Excel project that actually was released as an Office add-in for Excel a year ago as part of the Garage project. However, given the push to experiment across a range of features, Microsoft updated AFE and shifted its code into Excel Labs so users can avoid having to install multiple add-ins.
The tool lets developers more easily write, edit, and reuse complex formulas and LAMBDA functions, which were added to Excel more than a year ago. Microsoft at the time said that LAMBDA in Excel – a LAMBDA function is a block of capabilities that developers can pass around and use in their code – expands what spreadsheets can do and how users can write and share named formulas, but the Excel Labs functions should make it more efficient and creative, Microsoft claims.
"While the built-in Excel Name Manager lets you name and create complex formulas, the Advanced Formula Environment adds to that experience by providing capabilities typically found in modern code editors, such as IntelliSense, commenting, inline errors, auto formatting, and code collapse," Ronald wrote.
It's seemingly always about the AI
The second project – unsurprisingly for AI-happy Microsoft – involves artificial intelligence. The LABS.GENERATIVEAI custom function is designed to help enterprises determine how Excel would benefit from such models. LABS.GENERATIVEAI lets organizations display in Excel responses of OpenAI's large language models (LLMs) – think ChatGPT, GPT-4, and Dall-E 2— to queries.
Microsoft is investing billions of dollars in OpenAI and integrating its AI software increasingly throughout its own portfolio. Through this add-in, users can test the benefits of generative AI by sending prompts from the Excel grid to a LLM, with the responses being sent back to the worksheet. They can then run such tasks as parsing keywords from the responses.
- Microsoft boffins contemplate equipping Excel with AI
- Microsoft closes another door to attackers by blocking Excel XLL files from the internet
- Microsoft's new AI BingBot berates users and can't get its facts straight
- Excel's comedy of errors needs a new script, not new scripting
"The opportunities for using generative AI are endless as it can generate responses in a variety of different formats," Ronald wrote. "For the LABS.GENERATIVEAI custom function, we are specifically connecting to the OpenAI API for large language models."
Those models include "gpt-3.5-turbo," which powers ChatGPT, and "text-davinci-003" and are aimed at such tasks as text generation, transformation, and completion, as well as summarization and classification.
Organizations can use LABS.GENERATIVEAI to parse out keywords in responses to prompts and analyze the sentiments in something like a table of tweets.
Users can select LABS.GENERATIVEAI in the Excel Labs feature gallery by adding their OpenAI API key in the task pane. They then can enter =LABS.GENERATIVEAI and their prompt as the input to the function. As with OpenAI's Playground, they also can adjust the settings of the LLM's responses, such as "temperature" (for response the consistency of returned responses), and "maximum output length" (to control the max number of tokens returned in the response). ®