Machine learning and artificial intelligence in messaging will become commonplace.
Who would have thought that the most personal and manual form of interaction between humans can be mechanized? Years a go, it started with presence and instant messaging. People found out ways to communicate other than the phone call. Today, messaging is so prevalent that you have to take it seriously:
- In the consumer space, we’re talking about a billion users for these platforms. WhatsApp at 900 million is the closest to reach its first billion soon enough
- In the enterprise space, a single hiccup of Slack yesterday, sending many to vent off on Twitter
What is interesting, is how artificial intelligence is starting to find a home in messaging apps – consumer or enterprise ones – and where this all is headed.
I couldn’t care less at this moment if the interface is textual or speech driven. I might cover this in a later article, but for now, let’s just assume this is the means to an end.
Here are a few examples of what artificial intelligence in messaging really means:
The Silent Administrator
You are in a conversation with a friend. Chatting along, discussing that restaurant you want to go to. You end up deciding to meet there next week for lunch.
I do this once a month with my buddies from school. We meet for lunch together, talking about nothing and everything at the same time. For me, this conversation takes place on WhatsApp and ends up as an event on my Google Calendar.
Wouldn’t it be nice to have that event created auto-magically just because I’ve agreed with my friends on the date, time and place of this lunch?
This isn’t as far fetched as it seems – Google is already doing similar stuff in Google Now:
- Prodding me when the time comes to start the commute to a meeting
- Tracking flight delays when it finds an itinerary in my Gmail
- Giving me the weather forecast on mornings, and indicating “drastic” weather changes the night before
- Providing multiple time zones when I travel
Google Now is currently connecting to apps on the phone through its Google Now on Tap, giving it smarts over a larger portion of our activities on our phones.
Why shouldn’t it connect to Hangouts or any other messaging service scouring it for action items to take for me? Be my trusted silent administrator in the back.
A few years ago, a startup here in Israel, whose name I fail to remember, tried doing something similar to the phone call – get you on a call, then serve ads based on what is being said. Ads here are supposed to be contextual and very relevant to what it is you are looking for. I think this is happening sans ads – by giving me directly what I need from my own conversations, the utility of these messaging services grows. With a billion users to tap to, this can be monetized in other means (such as revenue sharing with service providers that get promoted/used via conversations – booking an Uber taxi or a restaurant table should be the obvious examples).
In the enterprise space, the best example is the Slackbot, which can automate interactions on Slack for a user. No wonder they are beefing up their machine learning and data science teams around it.
Knowledge base Connectivity
That “chat with us” button/widget that gets embedded into enterprise websites, connecting users with agents? Is it really meant to connect you to a live agent?
When you interact with a company through such a widget, you sometimes interact today with a bot. An automated type of an “answering machine” texting you back. It reduces the load on the live agents and enables greater scalability.
This bot isn’t only used to collect information – it can also be used to offer answers –Â by scouring the website for you, indexing and searching knowledge bases or from past interactions the live agent had with other users.
I recently did a seminar to a large company in the contact center space. There was a rather strong statement made there – that the IVR of the future will replace the human agents completely, offering people the answers and support they need. This is achieved by artificial intelligence. And in a way, is part of the future of messaging.
Speaking with Brands
If you take the previous alternative and enhance it a bit, the future of messaging may lie with us talking to brands from it.
As messaging apps are becoming platforms, ones where brands and developers can connect to the user base and interact with them – we are bound to see this turning into yet another channel in our path towards omnichannel interactions with customers. The beauty of this channel is its ability to automate far better than all the rest – it is designed and built in a way that makes it easier to achieve.
Due to the need to scale this, brands will opt for automation – artificial intelligence used for these interactions, as opposed to putting “humans on the line”.
This can enable an airliner to sell their flight tickers through a messaging service and continue the conversation around that flight plans with the customer throughout the experience – all within the same context.
The Virtual Assistant / Concierge
Siri? Cortana? Facebook M? Google Search?
These are all geared towards answering a question. You voice your needs. And they go searching for an answer.
These virtual assistants, as well as many other such assistants cropping up from start ups, can find a home inside messaging platforms – this is where we chat and voice are requests anyway, so why not do these interactions there?
Today they are mostly separated as they come from the operating system vendors. For Facebook, though, Facebook M, their concierge service, Messenger is the tool  of choice to deliver the service. It is easy to see how this gets wrapped into the largest messaging platforms as an additional capability – one that will grow and improve with time.
Why is this important?
Artificial Intelligence is becoming cool again. Google just open sourced their machine learning project called TensorFlow. Three days go by, and Microsoft answers with an open source project of its own – DMTK (Microsoft Distributed Machine Learning Toolkit). Newspapers are experimenting with machine written news articles.
Messaging platforms have shown us the way both in the consumer market and in the enterprise. They are already integration decision engines and proactive components and bots. The next step is machine learning and from there the road to artificial intelligence in messaging isn’t a long one.