Microsoft promises a new kind of shared interactive experience inside documents. But what does that really mean, and what does SharePoint have to do with it?
ZDNet’s Mary Jo Foley handicaps Microsoft’s vision to lead with its three core clouds: Azure, Microsoft 365 and gaming. Here’s the plan and other goodies developers will get.
When Microsoft CEO Satya Nadella talked at the Build conference about bringing “the next generation of real-time collaboration to the web”, the obvious comparison was Google Docs, which is still seen as the high point for co-authoring despite Office having much the same co-authoring tools both in the desktop and web applications.
The way the Fluid Framework breaks up documents into building blocks that you can collaborate on and combine into a new kind of document that isn’t limited to being a spreadsheet, a slide deck or document, may have rung a few other bells if you’ve been following the technology industry for long enough.
The idea of compound documents made up of chunks of information from, say, a spreadsheet and a charting tool and a word processor, sounds rather like what Apple was hoping to achieve with OpenDoc, or the ‘universal canvas’ of Netdocs, a productivity tool that Microsoft was building using XML to put multiple tools into a single early SaaS application. It’s also what you can already do with OLE compound documents in Windows, although it’s not widely used and performance has tended to be poor.
The near-instant appearance of text typed by one person on the screen of someone else working in the same document in the Fluid demos was definitely faster than the co-authoring you see in Word today, where content tends to appear in chunks rather than one letter at a time. What Nadella didn’t say on stage was that the cloud platform behind the Fluid Framework is actually SharePoint Online, and that’s where the impressive performance comes from.
The architecture for how SharePoint Online uses Azure SQL and Azure Blob Storage, and the way it talks to Office applications that open documents stored there has been significantly rewritten over the last couple of years, with a whole new file I/O stack for Windows based on the HTTP and IP protocols. The first place you’ll see that appear is in the Office Pro Plus applications, with PowerPoint not waiting until a large slide deck has downloaded before letting you start viewing or editing the first few slides, for example.
That new SharePoint storage layer on Azure is also what underlies Fluid, says Office 365 product marketing general manager Seth Patton. “Fluid is all based on pretty significant advancements that we’ve made in the turbocharged SharePoint file storage; there’s a bunch of back-end improvements that we’ve made that has resulted in everything from better file upload and download times, and Fluid is taking advantage of that for new collaboration scenarios.”
Patton describes the Fluid framework as “A new distributed data structure platform that allows for hyper-performant scenarios with AI included. Think about it as the ability to have, say, simultaneously 18 different people that are around the world in different geographies with not just real-time collaboration, but AI translations happening at the same time in sub milliseconds.” In other words, don’t think of SharePoint as slow or clunky, or just an intranet site and document library: think of it as “a new hyper-fast and performant cloud platform that has AI built into it.”
What you work with through that SharePoint storage layer and distributed data structure isn’t just a standard Office document; it’s an Office document broken up into pieces — “components that can then be shared across other apps that have the ability to collaborate within the end points [with the changes] coming back to the original file.”
So a ‘compound’ Word document might include a component that’s a table someone can be editing in the Word document, but that can also be shared into a Teams conversation where someone else can be adding more information. And if you turn the table into a chart and email it to more people, they don’t get a picture of the chart — they get a live chart that updates as the figures in the underlying table change.
Smart and shredded
This adds up to what Office 365 corporate vice president Jeff Teper calls “real-time, near zero-latency co-authoring on components between humans and AI.”
AI in Fluid doesn’t mean your documents getting written for you; it means a variety of tools to help you work, including the way documents get broken up into components. The real-time translation in the Fluid demo is one use of AI, but Fluid could have many more AI tools, from spell-checking to ensuring you’re not using internal codenames for products in a public document, to looking for personally identifiable information and other compliance issues.
The Fluid demo where an AI tool tip shows the best kind of chart to explain the data in a table will have looked familiar if you’ve already used Excel Ideas, which brings some of the Power BI automatic insights to tables in your spreadsheet. (It will still rely on having structured data with the right labels, but it looks like a more integrated experience.)
Think of those AI tools being like bots in Slack or Teams; you choose the ones that are useful and they run automatically in the background while people are working on a document or contributing to it through Teams.
AI will also help break documents up into chunks. One of the problems with compound documents before is that it’s been a lot of work for users to mark up and share chunks of documents other than by just copying and pasting them, which loses the connection to the original document.
Fluid will work with Microsoft Search to suggest useful content to include, building on the way Word lets you mention people with an @ sign and create to-do items inside a document to assign to them, or the way Bing can highlight key information from relevant web pages right in the search results, pulling out the relevant pieces of documents. Teper talks about what you can do with Fluid in terms of composable documents and discoverable components.
“One of the things we’re working on with Microsoft Search is shredding the content for pictures and slides,” he explains. “Going one step further, you’ll be able to say ‘insert quarterly sales’ and we won’t insert the quarterly sales slide presentation — we’ll have hopefully found the right table from the most-visited slide or document and suggest that for you. That relates to the Fluid framework as well, where we get more crisp about these components, so you can say ‘I don’t want the whole document, I just want this table’.”
At Build, Microsoft emphasised that Fluid will work across browsers; the demos showed tables and charts being edited and updated in the online versions of Word, Teams and Outlook. At the recent SharePoint conference, Microsoft demonstrated Fluid running in a SharePoint web page. “That’s a SharePoint web part with the SharePoint framework user experience layer talking to the new data layer, the Fluid layer, and persisting in to the new SharePoint storage layer built inside Azure,” Teper explained in the keynote.
But Fluid seems likely to work with desktop and mobile applications as well (and the SDK due later this year will let developers built it into their own apps). “We’ll bring this technology to multiple apps including — hopefully — eventually OneNote,” Teper told us.
There might also be brand new Fluid apps too. Doing the demos in Word was a way of introducing new ideas in a familiar context, but Teper pointed out the freedom of creating a new app like Teams where people don’t have so many expectations about how the software is supposed to work. “We’ll look at Fluid in experiences like that where we have a licence to create new muscle memory. We’ll use the technology in both evolutionary experiences, where [people will say] ‘great, it’s faster’, and in more revolutionary experiences.”
And if Fluid reminds you of ideas like the universal canvas that have been tried before, Teper wouldn’t disagree. “I think we have new technology to solve the problem better than it’s been solved before — including by us.”