Practically 5 many years after the launch of VisiCalc, AI is reshaping one of many world’s most acquainted productiveness instruments — the common-or-garden spreadsheet.
Whereas a whole lot of information employees work together with spreadsheets regularly, many lack the talents and confidence to entry extra superior capabilities.
“For a really very long time, spreadsheets compelled you to study spreadsheet syntax and spreadsheet methods of working,” stated Eric Birnbaum, director of product administration for Google Sheets. “However take into consideration how many individuals want to make use of spreadsheets at work and don’t have the talent set to create the sorts of spreadsheets that may be actually useful for them.”
The addition of synthetic intelligence might help deal with that problem, he stated, making the software program extra accessible to a variety of workplace employees. “AI is unlocking the facility of spreadsheets, taking up a whole lot of the troublesome work that’s required to make use of them. That may be extremely empowering for customers,” stated Birnbaum.
Google has steadily expanded generative AI (genAI) capabilities in Sheets since launching Duet AI — now Gemini — for Workspace in 2023.
Gemini in Sheets is offered at no additional price to Google Workspace subscribers, although a paid AI Expanded Entry add-on – costing $30 per consumer every month – is required to take away sure utilization limits.
Options which have rolled out in latest months embody the power for a Gemini agent in Sheets to hold out multi-step actions comparable to formatting, evaluation and knowledge entry, and, extra not too long ago, the power to create whole spreadsheets from a single immediate. A Fill with Gemini characteristic builds on the current AI operate, enabling customers to mechanically populate chosen cells by detecting intent from info inside a spreadsheet in addition to from the net.
One other characteristic, Sheets Canvas (at present out there in alpha), lets customers generate interactive apps that replace in real-time based mostly on modifications to spreadsheet knowledge. This may very well be a kanban board for a gross sales pipeline, for example, or an analytics dashboard that makes use of Sheets as its back-end knowledge supply.
Google claims customers already see a variety of advantages from Gemini in Sheets. In line with an August 2025 survey of 200 Sheets customers carried out by the corporate, the vast majority of information employees (89%) stated AI options in Sheets save them at the least an hour per week, and 88% imagine AI options have made them extra assured of their knowledge evaluation abilities.
The preferred AI use instances embody creating spreadsheets and charts, analyzing knowledge, and fixing damaged formulation. How broadly these options are literally used is unclear; Google declined to supply weekly utilization statistics for Gemini in Sheets.
As the corporate embeds Gemini deeper into Sheets, questions stay about simply how a lot companies can belief AI instruments to deal with necessary enterprise knowledge, in addition to what elevated automation means for individuals who spend a lot of their day wrangling knowledge.
Eric Birnbaum, director of product administration for Google Sheets.
Computerworld not too long ago talked with Birnbaum concerning the potential advantages and challenges of the newest evolution of spreadsheet software program. This interview has been condensed and edited for readability.
As companies develop into extra centered on seeing worth from AI investments, what measurable advantages are prospects seeing from Gemini in Sheets – for instance, time saved or different types of return on funding? “Spreadsheets stay one of many common languages for companies, and we don’t anticipate that’s going to vary anytime quickly. However by bringing AI into the product the place folks work, we predict it could considerably scale back the technical tax of information: the time spent understanding it, sourcing it, analyzing it, visualizing it.
“Traditionally, knowledge professionals spent — let’s estimate it at 80% of their time — on the mechanical groundwork, knowledge cleansing, crafting formulation, formatting, troubleshooting, and perhaps solely 20% of their time on precise strategic decision-making.
“We predict that by offloading the guide, time-consuming, error-prone knowledge work to Gemini, we will flip that ratio a bit, so people can concentrate on the issues that actually matter: the high-value inquiries to ask and the judgment calls and selections that get produced from them. We’re beginning to see proof of that in our personal consumer base and buyer base.
“The second level to make is that AI can democratize knowledge evaluation to some extent, and make it out there to many extra customers. For thus a few years, the spreadsheet was a gatekeeper; when you couldn’t communicate the inflexible language of spreadsheet formulation, you couldn’t extract the worth from knowledge. However by introducing these pure language interfaces, AI is separating the analytical functionality from the technical literacy.
“If you happen to can ask the best questions, or describe what you need in plain language, Gemini and Sheets might help you obtain your targets in a spreadsheet, even when you’ve got minimal spreadsheet abilities your self.
“What we’ve seen so removed from customers and prospects is that it’s extremely empowering for individuals who might need been scared off by knowledge evaluation or dreaded opening spreadsheets prior to now. You don’t have to go and anticipate an information analyst that can assist you; you’ll be able to go and do that work your self in a spreadsheet.”
The flip facet is, how assured can companies be if extra junior workers can tackle higher-level evaluation duties by counting on AI? Given the propensity for AI fashions to hallucinate, to what diploma can companies belief that these instruments received’t introduce errors into necessary enterprise knowledge? “It’s one thing we spent a ton of time interested by. We’ve gone to nice lengths to construct these AI instruments to collaborate with you, to point out their work, clarify what they did, and just remember to can take over the place they go away off.
“Our Sheets agent, for instance, lays out a very specific, clear plan so that you can evaluation and approve earlier than any knowledge manipulation occurs. It’s designed to do this in plain pure language in a means that the common consumer might perceive, after which it offers you again that closing abstract, so precisely what it did and the place it did it.
“The opposite factor is that the mannequin is nice at explaining issues. If you happen to inherit a spreadsheet that has some complicated method that you just don’t perceive, for instance, the mannequin does a tremendous job of explaining the way it works and what it’s doing.
“In some ways AI is just not solely making these options extra accessible, however serving to customers really feel extra assured within the output. Human error is an inherent danger in guide knowledge administration, with or with out AI. One misplaced comma or damaged cell reference can fully corrupt a whole monetary mannequin, and it may be fully undetected.
“Our method with AI in Sheets is to create this deliberate verification loop. You now have one other spreadsheet skilled working together with you, lowering the probability of those errors.”
Even when people produce errors too, does it finally come all the way down to accountability when AI is concerned? “Our viewpoint right here is that AI must be partnering with the information employee who’s doing the work right here. And every thing that we’ve constructed is designed to be that accomplice. You may be capable to offload duties to the mannequin, however we’re citing sources, we’re offering plans and explanations. We’re finally counting on the consumer to do this closing verification.
“We spend a humongous period of time centered on high quality. We all know that for AI to be helpful in spreadsheets, it needs to be dependable. Once we launched Sheets Gemini Agent, for instance, we have been actually proud that we set a state-of-the-art benchmark on the total SpreadsheetBench knowledge set, which on the time exceeded opponents and near-human skilled means.
“However we all know high quality isn’t ‘adequate’ or achieved. We’re continuously working to enhance high quality for our customers and prospects, and for the use instances the place they’re counting on AI most.”
What potential do you see for extra agentic performance in Gemini Sheets — for instance, bringing in knowledge from different sources, creating recurring studies, or taking extra actions independently? “We’re listening intently to prospects and customers and constructing what they’re telling us they want. The agent is already able to doing very complicated multistep workflows, and we see customers uncover that the agent is extraordinarily able to doing end-to-end spreadsheet duties. When it comes to connectors, in an alpha we now have connections out there to HubSpot, Salesforce, and Mailchimp. We hope to increase that over time.
“There’s no path to have AI changing analysts. I feel AI is giving analysts extra time again to really do the extra worthwhile components of their job. Analysts that I work with are far more productive and impactful than they ever have been earlier than, as a result of they will push that uninteresting spreadsheet grunt work off to the mannequin, liberating up time for extra fascinating and impactful work.
“A fantastic instance: the visualizations I’m returning from analysts these days are canvases as an alternative of static charts that I can discover myself. It’s far more informative and helpful than what I used to be accustomed to earlier than.”
Wanting forward, do you anticipate a bigger share of spreadsheet work to be carried out by brokers, with people setting targets and reviewing outcomes? What would be the largest change in how folks use spreadsheets with AI? “The duties are more likely to keep related and the use instances for spreadsheets are more likely to proceed to be related. The most important change would be the means to push the uninteresting spreadsheet grunt work off to a mannequin and release time for the consumer to do issues which might be extra impactful, extra fascinating, extra significant to the enterprise.
‘Spreadsheets have traditionally been these like static containers the place knowledge goes to relaxation. I feel AI can flip spreadsheets into these dynamic, localized software program purposes. And I do really suppose this kind of modifications the sport for a way folks may use spreadsheets transferring ahead.
“It’s not only a passive grid filled with numbers; spreadsheets are evolving to develop into these dwell, long-lived, collaborative purposes. Workers can construct these on the fly, like a fundamental CRM or provide chain dashboard in seconds. That is an space of funding for us transferring ahead, and we’re actually excited to see how this evolves.”
AI assistants and brokers, comparable to ChatGPT or Claude, may be capable to analyze spreadsheet recordsdata and enterprise knowledge with out customers working inside a spreadsheet utility. What do you suppose will preserve Sheets central to the workflow, quite than merely making it one a part of a wider AI-driven course of?“We’re in fixed contact with prospects and customers; it’s clear work continues to be occurring in spreadsheets. I feel spreadsheets stay extremely in style instruments. If we will carry the AI capabilities customers want straight into the product the place they already are, we’ll rework the best way they work.
“I feel we could be the front-end for a few of this nice AI innovation and the merchandise that customers are accustomed to at the moment. All the things we construct is guided by consumer suggestions: customers are telling us proper now they need AI to assist them do their on a regular basis or extra complicated duties, they usually’re beginning in Sheets at the moment.”
