A practical guide to taking advantage of AI in your enterprise, starting right now
Enterprises need to take advantage of new technology to stay competitive and relevant. If you don’t do it, someone else will. Before you know it, you’ll be out of touch with your customers, and they’ll move to your faster, cheaper, and more stylish competitors.
After years of anticipation, artificial intelligence is finally making a big impact in the business world. However, for most CEOs and business leaders, it’s still not clear exactly what AI is and how it can be used effectively.
By the end of this article, you’ll not only have a great understanding what AI means (and doesn’t mean), but also know how enterprises are already using it, and how you can too – starting today.
- Introduction to AI in the Business World
- AI Applications for Enterprises
- Figuring out What’s best for your business
A No-nonsense Intro to Artificial Intelligence
The most basic definition of Artificial intelligence is “computers acting intelligently” – performing tasks we usually assign to people, like understanding language, recognizing objects in photos, learning, and planning.
It’s not a new idea, the first AI programs were written in the 1950s. So why all the hype now? Well, up until now, the hardware and software needed to make machines act intelligently just didn’t exist. However, over the last few years, a few key advances have really opened the door for AI.
Big Data, Machine Learning, Cloud Computing – the Perfect Storm
Intelligent computers don’t really think, they just appear to. When Siri listens to your questions and gives you back useful answers, she doesn’t actually understand what you’ve said. Underneath is a computer program built from 1s and 0s just like every program ever written. However, the engineers at Apple haven’t coded up clever responses to every possible question. That would be impossible. Instead, they’ve built software that can learn from billions of past searches and questions, and take a good guess at what you’re looking for.
The three key ingredients to building an intelligent system like this are:
- Computer algorithms that can learn
- Data – a LOT of data
- A huge amount of computing power
Machine Learning Algorithms
Traditional computer programming involves coding repeatable tasks that a computer can execute quickly. Solving equations, scanning documents, and sorting files are all things computers have been great at for decades. However, when it comes to more complicated tasks like recognizing speech, this type of programming is too much work.
A better way is to create a program that can learn by itself, and then let it teach itself how to do tasks. This sounds crazy, but computers have been able to do this for a long time. The reason these programs haven’t been very good is they haven’t had enough data to learn from. Until now.
In the early days of the internet, content was created by companies and vendors. Now, everyone is generating information and data all day long. Social media activity, likes, photos, GPS locations, online purchase histories, and so much more are being recorded every second. Every day we create 2.5 quintillion bytes of data – that’s over 90 years of HD video – and 90% of the world’s data was created in the last two years.
This tidal wave of information can be used to train the self-learning programs described above. Using real world information, these algorithms can learn about the way the world works, and predict what might happen next.
The final piece of the puzzle is raw computing power. Showing a face recognition algorithm billions of photos to learn from is no small task. To do it you need access to thousands of computers to store and process all the data. Until now, only big companies like IBM and Google had access to resources on this scale.
Download Our Project Specification Template
Cloud computing has changed all this. Computer storage and processing power are now offered as a service from the likes of Amazon and Microsoft. This has brought large-scale computing power to everyone, at an affordable price.
Put these together and you have all the ingredients to build intelligent systems.
Putting Them Together – What’s the Result?
Organizations all over the world are using all of the above right now. They are using cloud IT infrastructure to store run machine learning algorithms on all kinds of data. But, what are they actually building?
Here’s a quick list of things AI is already doing:
- Speech recognition – Siri, Cortana
- Research – e.g. discovery at legal firms
- Consumer behavior analysis – predicting what you want to buy next
- Fraud detection – spotting suspicious behavior
- Network and IT security monitoring – when might your system go down
- Market projection – what will people want?
- Sales forecasting – how much will they buy?
- Office automation – writing those boring reports for you
…the list goes on, and it’s growing every week.
Common AI Misconceptions
So are computers going to take everyone’s jobs? Will robots take over the world?
No. At least not in the foreseeable future. That is AI general intelligence, where a computer is as smart – or smarter – than a human at any task it is given. Don’t worry, we aren’t even close to that yet.
The type of intelligence we are talking about is AI narrow intelligence. These are systems that are great (usually even better than people) at one specific task. You can think of them as tools that are smart but only in extremely specific circumstances. We’ll see what these things are a bit later.
Why it’s Important
People are already becoming accustomed to this technology. The service from companies like Google, Uber, and Amazon is fast, accurate, and personalized. This quickly becoming the expectation for all companies. Shoppers don’t want to spend valuable seconds searching for related products, they want instant recommendations. This type of service can only be offered by AI.
Beyond customer service, AI and data analytics are becoming important tools in keeping businesses running safely. In a hyper-connected world, your IT staff don’t have a hope of defending your company against sophisticated attacks without powerful tools.
Check out the results of this survey on what business executives think are the most important benefits of AI.
AI isn’t new anymore. The enterprises that adopted it early are already reaping the benefits. Now, it’s a case of who is going to get left behind.
Who’s Using it?
The quick answer? Pretty much every tech company, and a good number of others. Some easy examples are:
- Facebook – Image recognition, page recommendations
- Google – In their search engine
- Amazon – Marketing, sales forecasting
- Microsoft – Customer service
- Tesla – Driving their autonomous cars
- IBM – IT security (more on this later)
- New York Times – To write routine articles like financial summaries and sports reports
- Apple – Siri
Even smaller companies are getting in on the game. One survey found that 38% of enterprises are already using AI, and 62% will be by 2018.
So How Can You Get in on the Action?
So far we’ve got an idea of what AI is, what it’s not, why it’s important, and who’s using it. Now, let’s take a look at some real applications you can start using right away to bring the latest tech to your company.
AI Applications for Enterprises
Ok, you’re convinced. AI is worth thinking about, and you’re ready to look into uses for your business. So what’s available? What products can you get your hands on?
In this section, I’ll be going how to implement AI in your business, and some great places to get started.
Summary of Applications
Feel free to click the links to jump to the sections most relevant to you.
AI in Sales Automation
For most businesses, that brief moment when a customer makes a buying decision is vitally important. That’s when you want to be fast, accurate, and ready. You need to know what they want to hear when they want to hear it.
Using AI and data analytics, you don’t have to guess anymore or generalize for all your customers. You can deliver the right recommendations, special offers, and incentives to close more sales on a predictable basis.
Example: Salesforce offers some great products to get you started with this sort of tech. Their Einstein platform has some cool features such as:
- Predictive Lead Scoring – Automatically get a score for each of your sales leads, letting you know the likelihood it will convert into an opportunity.
- Forecasting – Forecasting results based on data models, letting you know the likely value of something ahead of time.
- Recommendations – Recommending the right products to customers at the right time. You’ve seen this before with Amazon product recommendations. But, this can be extended to all sorts of things, like personalizing your email campaigns.
Build AI Right Into Your Apps
Web and mobile users are getting used to some of the amazing apps currently available. Apps from Apple, Google, and Facebook have some awesome features like face and voice recognition.
Read How We Helped a Marketing Company to Build a Back-Office Custom Ads Dashboard
If you think these types of features are out of reach for your company’s apps, you’re wrong. Now you can access this type of Artificial Intelligence as a service. What’s better is that you don’t need any specific knowledge of AI or how anything works. Just plugin to an API and you’re good to go.
Some of the things you can do are:
- Speech Recognition – Google Speech API lets you tap into the same technology that powers Google’s own products. It can convert speech from a device’s microphone or an audio file and can understand over 80 languages.
- Image recognition – There are a million cool ways you could use recognition in an application. APIs like Clarifai, Imagga, and Google Vision API allow you to easily recognize objects, faces, and text in images. You can even do things like search images based on object similarity.
- Video analysis – Training a good video analysis bot requires a ton of video content to train it on. That’s why it’s no surprise that Google – with all the content on youtube – has a great video intelligence API. It can watch a video and tag each frame with what’s in it and where the shots were taken. Valossa attempts to go further and add things like emotion into its analysis.
Upgrade Your Security With Machine Learning
Every company, individual, and device is now connected to the world in ways that no expert or traditional software can understand. These days, fraud and cyber crimes are sophisticated, targeted, and ever changing. Having a firewall or traditional antivirus software installed is no longer adequate. Especially not for an enterprise.
The smart guys are now using AI tech to finally start winning the battle against cyber criminals. There are already some awesome tools on the market, both from large companies and startups.
IBM is using ‘Cognitive Computing’ (their term for Artificial Intelligence) to build their new generation of IT security products. Their famous IBM Watson can learn from security data that other systems simply can’t. 80% of security data on the internet is unstructured and designed to be read by humans. However, IT security professionals and data scientists are expensive, and even the experts can’t read everything. Watson can read and interpret all of this information.
What’s more is Watson is a single security system. That means that any threat, data, or insight found around the world will be instantly available to anyone using the system – in real time. It’s like having an expert that is constantly learning from everything at once.
One cool startup called Fraugster has been gaining some serious momentum in fraud detection. Ther AI software goes beyond traditional rule-based systems and can learn from a sea of unstructured data from different sources. In their own words:
“We have invented a self-learning algorithm that mimics the thought process of a human analyst, but with the scalability of a machine, and gives decisions in as little as 15 milliseconds”.
Their AI uses a crazy number of data points. Things like name, email, billing and shipping address, IP latency check to find out the real distance to the user, IP connection type, distance between keystrokes, and email name match to find out if a customer is legitimate.
Improve Your Customer Service With a Chatbot
Another great place to begin implementing AI solutions into your enterprise is your customer service. Whether you’ve noticed or not, you’ve probably talked to an AI chat bot before. Companies of all kinds have started using them to serve customers more quickly and accurately.
Probably the best product available now is the Microsoft Bot Framework. It’s a framework you can use to build a chatbot for your own company that talks and interacts with your customers the way you want it to. You can check out the details of how to build a chatbot on their blog.
The cool thing about the Microsoft Bot Framework is that it goes far beyond just sending predetermined answers to questions. They have trained the AI with Microsoft’s huge resources and data from their Bing search engine. The AI chatbots can do things like:
- Solve real problems for your customers
- Authentication for customers
- Call other bots using Skype
- Recognizing images
- Interpret what customers actually mean, rather than just looking at the words they say
I think the coolest part is that the bots can actually understand customer sentiment, and have emotional intelligence. The level of language understanding is so high your chatbot will be able to tell if a customer is happy, annoyed, frustrated etc. and hand over to a human agent when necessary.
The easiest way to start using automated customer service is as a supplement to your current system. For example, bots could help out with simpler problems, and your trained human agents could handle the more difficult tasks. Or, chatbots could be used to help with unexpected surges in support demand. This could be when one of your services goes down, and many people are asking similar questions.
If you don’t want bots actually talking to your customers, there are other options. Some AIs can listen to past conversations with customers to determine their needs and personality type. Then, use this information to match them with a suitable customer service rep they are likely to be happier with.
Automate Your Routine Business Processes
One of the more practical, but less exciting ways to bring AI to your organization is to automate your business processes. These could be any tasks any of your team perform in your office on a routine basis. Many of them are tasks you assume have to be done by a human.
This includes things like:
- Accounting – There is already software that helps with accounting, but cool startups like SMACC are helping companies reduce their accounting bills and headaches in new ways. Things like data entry, bookkeeping, and payments can be done faster and more accurately by leveraging AI.
- Basic Journalistic Work – Yes that’s right. AI bots are already writing articles all over the internet. Ok, they aren’t writing the most interesting articles, but they are writing things like financial reports and summarizing political events.
- Team-Project Matching – If your company regularly needs to match projects to a suitable team, AI can help with that too.
- HR Resume Matching – Recruitment is a notoriously difficult task even for the best talent scouts. Now services like IBM Watson Talent are using data and machine learning to help you decide who’s the best fit for a position.
- Legal Discovery – Bots have already started to do the work of junior lawyers.
The list goes on and on. However, these kinds of tasks will usually be unique to your business. Finding a dog in an image can be done with one API request, but finding the best developer for your project can’t. These types of tasks need a much higher level of customization.
To make AI and data analytics work for your business, you might build your own custom software – or find someone that can do it for you. I wrote an entire article on this topic a few months ago.
Go Big – Transform Your Industry
Finally, if you really want to get your businesses hands dirty with AI, you can partner with an AI firm to develop something truly unique. I’ve mentioned some of the ways you can use the IBM Watson APIs to get some cool results quickly. But, Watson is really aimed at much larger applications that transform industries.
Of course, this won’t come cheap or quickly. For these types of projects, companies like IBM will have PhDs and data scientists working specifically on your problem. Then any solution needs to be trained and tested, especially if you’re the first one to attempt something.
Figuring Out What’s Best for Your Business
The methods listed above is just a starting point. A truly comprehensive list of ways to start using AI and data science
Finding out which is the best way to start using AI technology in your enterprise is a tricky process. You’ll need to do a few things:
- Understand completely the systems you have right now
- Research in depth all of the methods above and how they could be implemented in your company
- Ask yourself – “What will this provide that I don’t have already?”
It’s tempting to dive right into the coolest sounding technology. But, the best way is to start with a business case that actually needs solving. You might start by researching companies similar to yours to see what they are already doing, and what has worked for them.
If you’re unsure, you’ll probably need an expert opinion. There are some great companies out there that specialize in helping business transition to using technology like AI, data analytics, and cloud computing.
Once you’ve decided the best way to start your enterprise with AI, you need to get to work. The first thing to do is work out your budget. This will have a huge impact on the way you can approach things. You can ask yourself:
- Which products are in my price range?
- What’s the price structure? And how will it change as we scale up?
- What kind of developers do I need?
- Can I use my current developers? How much training will they need?
- Do I need a data scientist?
- Shall I partner with an AI company?
AI isn’t an experimental technology anymore. The early adopters have researched, tried, and tested its business applications, and are using them right now. Within a few years, almost every company will be using data and AI to cut costs, improve their products, and make customers happy.
The methods I’ve outlined here are some of the best and most popular ways to introduce AI to an organization – whether it be a startup or a large company. If you don’t start taking advantage now, you’ll quickly find your margins shrinking, your headaches growing, and your customers moving to smarter competitors.