This year both Microsoft and Facebook have announced their commitments to bots and released development kits for organizations to develop their own bots for their respective messenger apps.
Microsoft for example has built a pizza ordering bot that you can have an informal chat with over Skype to complete your order.
Facebook has demonstrated a flower ordering bot from 1–800-FLOWERS where instead of pressing 1 for roses, 2 for tulips, etc. you could now swipe, click and type your way through the process. Supposedly this will take up less your attention than a phone conversation would, as you could frequently nip in and out of the chat with the bot.
These bots are starting the post-app era, where companies realize that getting people to download your app is actually quite hard and not always worth it. Through Natural Language Processing (NLP) it is possible to serve almost 100% of the use cases a bespoke app offers, simply by chatting with the user through a 3rd party messenger app.
But in the hype to make everything a bot now, what bots actually are what they aren’t isn’t always as clear. A bot really is a new kind of UI, a message, which gives it its name ‘conversational UI.’ Now unlike message in a 80s command line, a bot can use NLP to understand you and correspond with you like an intelligent agent.
So unlike conventional user interfaces (which have buttons and surfaces you manipulate through click, swipe or type) bots can be interacted with through messaging platforms like Facebook Messenger, Skype, Slack, email or even text messages. There are either one or two layers of AI at work to recognize your intent without relying on specific syntax. The first is Natural Language Processing (or short NLP) which uses AI to interpret our human fuzzy language and make a best bet on what it is you likely wanted from the machine. The second layer is optional and is Voice Recognition: Of course you might prefer a written conversation, but you might also want to speak with it (for example whenever you don’t have a free hand). And in that latter case there is another AI at work to decipher your soundwaves and transcribe them into text. So chatbots may seem like a simple piece of software to use, but only because the backend side of things is doing a lot of the heavy lifting for you.
At the workplace there are dozens of use cases for bots (always keep in mind the use cases below are not the ones you would, but that you could hand over to your bot assistant):
- Ask how many days of leave you have left
- Then subsequently request more leave which triggers the appropriate workflow in your backend
- Scheduling a meeting with your team by finding available slots between all your team members’ calendars
- Checking in with a Slack group whenever you arrive/leave a GPS location
- Searching for documents you recently worked on
- Asking colleagues what they want for lunch to conveniently collect their orders
- Reply to emails with short commands while the bots adds a salutation, ending a signature automatically
- Check in with team members to track daily activities
- Collect analytics data you frequently request
- Respond to workflows tasks quickly (for example by approving or declining them)
- Suggest a new restaurant go to for lunch, based on colleagues’ Facebook or Foursquare check-ins
- Pre-order Uber / Starbucks
- Research and book a flight
- Reminding you of deadlines you may have agreed to in emails to managers (and might have forgotten about)
- Alerting you to emails that seem important
- Search in XYZ data source
- Congratulating a colleague for their birthday or anniversary
- Randomly delight a colleague with a joke through email or Slack
- Direct a specific question to other users that might have knowledge in said area
- Check if a colleague is at their desk
These are just 20 more or less simply examples, which show the easy nature of bots: They can only do one thing but that in a very straightforward manner — they’re short of sweet. The list of examples is non-exhaustive since technology has only recently discovered bots.
But as they proliferate it becomes clear as the number of bots increases so does the need for an intelligent assistant that consolidates all of these bots and negotiates with them on your behalf. For example, many bots will need to know your name or location at some point during their task, but the assistant could provide such info right away without having to redirect the question to you.
Bots are a marvellous invention. While still in an early phase many experts agree that they have the chance to achieve what many big software programs have never been able to lay claim to: Making many little, annoying things a little less annoying.
