The bot revolution is upon us. Should we rejoice as the weight of tedious tasks is lifted off of our shoulders or be worried about these software applications and what they’ll be able to do? Should we even feel anything at all?
To answer that, we first need to understand what bots are, what they can do for us, and what is actually considered a “bot.”
What are workflows?
Workflows are static, rule-based tasks set up to save us time; take over a tedious, repeatable task; or even provide a fun diversion. These are more commonly known as Outlook rules or filters, IFTTT recipes (which can be fun or helpful), or even the old-school BMC’s Ctrl-M workload automation.
They can be a lot of things, but they are not bots, even though they are frequently named so.
Let’s take a look at a few workflow examples.
An IFTTT recipe might be designed to automatically log my work arrival and departure times into Google Drive, based on a specific location. The recipe does just that and doesn’t have other dependents, such as whether you arrive by car or bus, during the weekend or on a weekday, or whether you’re late or surprisingly early.
An Outlook workflow is based on a rule or a filter. When an email from John Smith arrives, the rule kicks in and files it in a “marketing” folder, or a message from a social network is marked as read, or an email in my inbox that is older than five weeks is deleted, and so on.
Workflows are amazing automation tools that help us in our jobs, but they are not bots. Unfortunately, the word “bot” has been inflated and hyped so much that many workflows are mistakenly referred to as bots. Just because a rule-based script was created to follow an action doesn’t make it a bot, just like we can’t call every Excel spreadsheet “big data,” every hosted service a cloud, every brushstroke on a canvas a piece of art, or every DJ a creative genius. (You can dispute any of these generalizations, but you get the point.)
How are bots different from workflows?
Bots attend to more complex issues and don’t necessarily follow one rule but may instead require numerous actions or steps to fulfill a task. In addition, they are able to adapt to a situation thanks to their ability to adjust to different states or requirements.
Consider Baymax from the movie “Big Hero 6” as a bot with a single task — to improve the well-being of his owner. He will fulfill this task through various means: a simple hug, flying in the air, or even inquiring about exterminating a villain. Note that he cannot deactivate until the task is completed or his owner is satisfied.
We can also compare workflows and bots in terms of how both can be used to generate reports. A workflow will complete the task per its specifications, yet it isn’t dynamic enough to make any changes. If, for example, you need a six-month report instead of the regular monthly report, the workflow itself is useless, whereas a bot can adapt to fulfill the task it is given.
Another way to look at it is that the workflow generates a specific, predetermined report, while the bot is capable of waiting for specifics each time you want a report.
Are all bots the same?
Clearly not! Bots differ by behavior as well as by the task they’re designed for.
There are delegation bots and contextual bots, but this is where the unfortunate misinterpretation begins. You see, delegation bots aren’t truly bots, but rather fancy workflows. These take center stage on your news feed, as chatbots within apps, and even as Slack bots, yet in their current state, bots are better without conversation. The A.I. is just not quite there yet.
Delegation bots are simple; contextual bots are more complex and are divided into autonomous and interactive bots.
1. Autonomous bots
As you might expect, these bots fulfill a task automatically to the best of their knowledge, yet they also require trust, as they make some decisions themselves.
When you request that an autonomous bot order a large pizza, it will do just that, and it will make its own decisions if queries arise that weren’t noted by you ahead of time. For example, it can decide whether the pizza should be thin or thick crust, include red or green pepper, etc.
Another example is autonomous driving cars. In order for the car (bot) to drive automatically, it must have control; bots have quicker instincts and response time and can make decisions in certain situations much more logically than us humans.
Of course, the technology is still progressing. Consider the Tesla accident that occurred recently. Without going into right or wrong decisions, the bot drove and reacted to the situation as required by its knowledge base to avoid a certain crash. In this particular situation, it made the appropriate decision to avoid one crash, though, unfortunately, a different one occurred.
2. Interactive bots
These bots need more info, or, as Johnny 5 from “Short Circuit” requested, “more input.”
This doesn’t work in a driving context, because if a car cuts you off, there is no time for questions about whether to slow, veer left or right, etc. But the pizza robot may fit this model better. Before it adds toppings or orders a crust you don’t like, it can ask what you want.
In a botshell
The potential of bots is immense and exciting. We have so much to look forward to. But this doesn’t mean we need to call every rule or automated action or workflow a bot just to serve up the hype. Workflows are great and can save you time, reduce mundane tasks, and take care of repeated jobs for you, but they won’t be driving you around or becoming true personal assistants, as bots will in the near future.