7 Founders Share Their Secret to Building a Successful AI App
The artificial intelligence and voice recognition space has been growing rapidly. According to a Gartner report, by 2020, 85% of customer interactions will be managed without a human. This is pretty much likely, as we are already teaching our machines to interpret data into logical solutions.
Apps running on artificial intelligence should make a user’s life easy, but what do application development companies like us do to build such apps?
To understand it further, we spoke to some successful founders who have built and scaled apps running on artificial intelligence algorithms.
Here’s what they have to say:
#1 Provide an experience that becomes a habit because it’s so valuable
Nathan Benaich, Investor in tech companies Playfair Capital, all things data, machine intelligence, user experiences
A truly great application powered by AI will deliver that wow moment, the magical feeling of accomplishing something that wasn’t quite possible before, right out of the box. Take Google Photos, which can retrieve images matching a search query. Or SwiftKey, which predicts the words you’ll type next. Or Tesla’s autopilot, capable of taking control from a driver. These products all elicit the feeling of pleasant surprise, an experience out of the ordinary that quickly becomes a habit because it’s so valuable.
#2 Your algorithmic approach should be invisible and irrelevant to the user
#3 Best AI completes a task start to finish for the user
Dennis R. Mortensen, CEO and Founder, x.ai
The best AI products make the user more capable. And the very best AI completes a task start to finish for the user; the user should have to do as little work as possible to make it happen. These are truly autonomous intelligent agents.
It’s just as important that these AI agents or virtual assistants allow users to communicate using natural language. Having to modify speech or text into some sort of machine syntax is cumbersome, and requires the user to do work. The whole point is to seamlessly hand over a task.
On the other end of the interaction, it’s important for the AI’s responses to seem as natural as possible. At x.ai, our goal is for our customers to feel as if they were interacting with a human assistant. And as we began to build our AI personal assistant Amy Ingram, we found that a collection of templates would not achieve this effect.
So we ended up creating an entirely new role– the AI Interaction Designer– who is tasked with developing Amy’s personality and ensuring she has a coherent voice. Our customers love Amy and how efficient she is, but what they remark upon most is how human she is. Many of the people they schedule with don’t even realize Amy is AI.
#4 The real power of AI is augmenting and enabling humans
Manuel Ebert, Founding Partner at summer.
Here’s what doesn’t make for a great AI app: taking something we already can do easily, like ordering burritos, and just slamming a chatbot on it. The real power of AI is augmenting and enabling humans, like an intellectual steam shovel.
A great AI app will not just help users do something faster, but enable them to do something they couldn’t do before. This includes things users couldn’t afford before: AI will empower all of us to have our personal travel agent, stylist, nutritionist, lawyer, coach, and financial adviser.
#5 In the best AI app, tech isn’t in your face but blended into the experience
Raj Singh, Co-Founder/CEO Tempo.ai, acquired by Salesforce
The best apps leveraging AI do it in a way such that tech isn’t in your face but blended into the experience.
I think Google Photos is a good application of serious computer vision tech but is presented in a delightful way where the tech isn’t getting in the way of the experience.
#6 Entrepreneurs building AI into their app should follow the 80/20 rule
Dr. Rob McInerney, Founder and CEO at Intelligent Layer
The greatest advice I can give to entrepreneurs looking to build AI into their app is to follow the 80/20 rule and understand how to manage uncertainty.
When applying AI techniques it’s easy to get stuck worrying about every single possible input, but in reality, the majority of the value to the user comes from solving a small subset of those inputs and doing that really well.
At Intelligent Layer we spend a lot of time decomposing user behaviour and identifying what the most common actions are so that we know where to spend our machine learning effort.
We also work very hard to manage uncertainty, which means that our algorithms know if they’re facing a part of the problem space they’re inexperienced in or not good at solving – they can then farm out to a human or revert to a set response. Ultimately this gives the user a far more satisfying and genuine experience.
#7 Exceed user requirements through constant interaction refinements
Snehal Shinde, CTO & Co-Founder at Mezi
A great AI app needs to have conscious and needs to emotionally connect and delight the user. It needs to make the user feel comfortable and act like a knowledgeable and thoughtful friend. It can only do so by learning about the user with every interaction and personalizing the experience accordingly. The best AI not only meets the user’s requirements but exceeds them through constant interaction refinements.
Predict what are the things that don’t exist right now but your users will like in your product
Xavier Amatriain, VP of engineering at Quora, ex-research/engineering director at Netflix, best known for his work on Machine Learning
In a #Bitesize interview with us, Xavier Amatriain explained the key to successful machine learning in developing products, here he explains what makes a great AI app:
Even just staring at the data and coming up with hypothesis are also going to help you figure out that what are the things that you could improve in your product and what are the things that can help in your business.
You can go to extreme as to even predict that what are the things that don’t even exist right now but your users will like.
That something Netflix is known to figure out – what is the next movie that they need to produce that doesn’t even exist right now. So, by looking at the data you can understand and you can come up with a hypothesis of things that even as of today don’t even exist.