3 innovations for smarter Intranets
For our all-new Now Assistant we have identified three key innovations that signal a new era in Intranets. They allow to break down silos, make your Intranet available literally everywhere and make communication with them more intuitive and natural. In this post we will explain these three innovations.
First up: New ways to Search
Search, while making an appearance in almost every single enterprise application, is seldom executed well – but even when it is, it is still very expensive and complex to use. Some applications have very extensive advanced searches, other are just a plain search through titles, some have full text Search, a few OCR text recognition search, etc. The big problem is that every Search is just a search within that particular silo, there almost never is a way to Search more than just that one source you currently use.
Of course there’s dedicated Enterprise Search, but that’s traditionally, for reasons too long to mention here, suffering from exceptionally poor user adoption.
In one example the marketing department daily searched through a whopping 12 different sources to get their daily measurements and indicators. No one in their right mind would search Yahoo, Google, Bing, let alone nine more search engines to look up a single fact.
Imagine you ask your run of the mill SharePoint a simple question: “How many days of annual leave do I have left?” What would your answer be? Unsurprisingly to some, the answer is no results are found.
So the obvious question is: Why is Enterprise Search not more like Google? It is in fact the million dollar question. Entire industry groups like AIIM are discussing the finest nuances about search year on year.
Now Assistant does things a little differently. As it turns out it is entirely possible to use the same technologies Google uses to makes its Search smart to make business search smart. This brings us to AI capabilities embedded into Search.
1) Voice Recognition
We all remember when Siri first came out it’s accuracy was kind of hit and miss, which is acceptable for its then Beta status. But since then not just Apple, but Google, Amazon, Hound and a few more have stepped up their game tremendously. Over the past 5 years reliable and fast voice recognition has emerged across many languages, platforms and devices. Without exception these systems rely on an active internet connection for a good reason: Algorithms that interpret sound waves and transcribe them into words use a lot of computer power. Too much actually to be done on the phone itself, it would just sap to much costly battery.
Developers today can tap into APIs to use the big firms’ prowess in recognizing voice to use it for their own applications. Some free, some paid. Google Chrome for example consistently offers a voice input feature for any HTML5 input field. Amazon is making Alexa available through an API, Wit.ai from Facebook is doing something similar that even works on iOS. In short, voice recognition is about to be commoditized and will find more standardized implementation in development frameworks.
Natural Language Processing (NLP) is coming from the idea to recognize natural – in other words human – search commands, e.g. to navigate your car’s head unit or search Google. But in recent years even major operating systems, like Mac OS X or Windows 10, have introduced NLP for their Searches. On a Mac for example I could search for “emails I received last week from Martin” and AI behind the search automatically recognizes which part of my query refers to the sender, the recipient, date, and so on…
NLP allows self-learning AIs to recognize the intent of the user and then execute the corresponding search for them. This technology is at the core of many AI API providers, like Wit.ai or Api.ai. Such services can help businesses provide Google-like instantaneous answers to their employees. Which brings us to the format in which such updates are delivered to the user.
Bringing updates: Card-based UIs
Let’s imagine you ask your (presumably now smart) Intranet a question: “How many days of annual leave do I have left?” – only this time the answer you receive isn’t zero search results, but a card.
That’s an interesting new angle to look at search. It’s not just anymore for pure finding of documents but to receive answers to actual questions. And Cards are a perfect vehicle to deliver such answers:
They have a short and definitive answer to the user's question. Below they show rich media or dashboards that visualize relevant data. Then you can see a colleague you could inbox with any questions/problems, along with relevant policies you might want to read up on. And finally, when you ask for remaining leave days, chances are you want to request leave – so the card lets you start a new request on it. Neat.
Cards are great because they show condensed information at a glance and are self-contained, meaning they don’t depend on any surrounding buttons or UI to work. This allows you to place many different cards (so with widely varying content) below each other, without risking to confuse your users. This also makes them perfect for mobile devices by the way, as you can show them in a stream or side-by-side depending on how much space you got. These benefits are responsible for the widespread adoption of cards in UIs of many consumer services, i.e. Google, Pinterest, Facebook, Twitter, to name a few.
So by the time users encounter their first card at work, they will already be familiar with how to use them as they’ve seen them for years on their Facebook…
Another drawback current compartmentalized UIs of most Intranets are suffering from is that everything lives in its own little box (called Portlet, WebPart or Widget, depending on the product). And you always see the same elements on the homepage, whether or not you actually need them. And in many cases they even show the same items, too. Chances are I’ve read the latest company news 5 days ago -- but current Intranets, you guessed it, have no way of knowing whether this item is still relevant to me.
Cards help with that. They are aware of a user’s role in the organizations, which audience they belong to, their colleagues, documents they or colleagues recently worked on, community pages they have joined, etc. And by harnessing and using all of this info, Cards can become a highly individualized personal stream for the user, that only shows info that’s uniquely new to the user. And when they’re done looking at something a card can be swiped away to be archived or dismissed.
Bots for conversations between human and data
Bots are a very popular topic since of very recently. They are yet a new kind of user interface: Messages. But unlike a command line from the 80s, the commands you send to a bot are interpreted with NLP (Natural Language Processing) which allows bots to understand and reply to you in a fashion that is similar to how a human would interact with you. There are also many different kinds of bots, each specialized in just one skill, i.e. flight reservation, e-commerce, leave request, etc.
Because bots all share the same UI – messages – they can be all integrated into one and the same messaging app. Techcrunch therefore predicts bots will replace apps and bot stores will be the new app stores. And indeed since Facebook and Microsoft both announced their commitment to bots and are bringing out bot development frameworks for developers, it is certain to say that bots are here to stay.
A leave request bot for example could be started by calling it from Skype for Business or Slack and asking it to fill out a leave request form for you. The bot then needs to know three things: How much leave you are entitled to take, how long you want leave for and what type of leave you request (parental, annual, jury duty, etc.) The entitlement it may take from a payroll software but the rest of the variables the bot will capture through an informal chat with you – the user.
Users presumably find bots easier to use for tasks they don’t perform everyday or in situations where chatting or speaking are the user's first choice. Since the bot can recognize the intent of the user from their initial message (i.e. “I want next week off”), it only has to ask those follow-up questions necessary to fulfill the request (“what type of leave do you want to take? a) Jury duty b) Parental leave c) Annual leave, etc.”).
The three innovations covered in this post are central to adenin’s all-new Now Assistant that enhances existing Intranets by adding data access to outside data sources and sending users proactive updates as well as offering NLP voice search.Click here to sign up for a live demo