As we look back at the past decade of innovation in the private and the public cloud space, led by Amazon, Microsoft, Google and IBM, the most significant emerging trend we see is the drive toward serverless computing and the appliance model.
In the initial days of cloud computing, companies used cloud as a substitute for their colocation facilities and/or data centers. There were certain incremental benefits to this approach. One benefit was moving capital expenditure away from an equipment model to an operational model. Another was arriving at a service model where the cloud providers themselves take care of software updates, which was especially true with companies like Microsoft and Oracle. If you were using Microsoft software, for example, you wouldn’t need to worry about the periodic operating system updates in managed instances of Windows server virtual machines.
As cloud computing has advanced, more companies have made the transition to the cloud-based platform as a service model (PAAS), which delivers computing and software tools over the internet. PaaS can be scaled up or down as needed, which reduces up-front costs and allows you to focus on developing software applications instead of dealing with hardware oriented tasks.
To support this shift toward the PaaS cloud, public cloud companies have begun heavily investing in building or acquiring serverless components that have pre-built unit functionality. These out-of-the-box tools allow organizations to test new concepts, iterate and evaluate without taking on high risk or expense. In the past, only large companies with considerable resources could afford to experiment with AI-based innovation. Now startups or small teams within larger enterprises have access to cloud-based, prepackaged algorithms offering different AI models that can fast-track innovation.
Let’s explore practical examples of how this trend helps democratize innovation in artificial intelligence by minimizing the time, money and resources needed to get started.
Resolving The Innovator’s Dilemma
Imagine your company makes kiosks or digital signage that is used by fast food chains. When customers pull up to a drive-through kiosk, they all receive the same menu choices. But what if the kiosk was smart enough to personalize recommendations? What if it could provide a suggested food and drink pairing based on the weather and the demographics of each customer?
If you wanted to investigate this idea several years ago, you would first write censor software to determine whether someone is within a certain distance of the kiosk. Then you would write censor software to detect weather information, followed by software for recognizing someone’s face and identifying their demographic information. Finally, you would write the program proposing food and drink options based on the gathered data.
The biggest challenge is that this process requires substantial time and money, and you have no guarantee that your idea will be viable in the end. Will the kiosk work as intended? Will the market be ready for it? Will your customers see value in it?
Now with cloud computing, you can explore your idea without a huge budget or team. The cloud facilitates innovation, not only from a technology standpoint, but also from a business, market validation and iteration perspective. The serverless public cloud infrastructure of all major providers – Microsoft, Amazon and Google – comes with ready-made components, like face-recognition tools and edge sensors that detect movement and weather conditions.
A software developer on your team could use these components to build a quick prototype and test it with a select group of potential customers for proof of concept. It would be feasible to create a minimum viable product in three to five months, roll it out to select locations, then use feedback to iterate on enhancements. If the concept doesn’t work out for any reason, your sunk cost would be significantly lower than in the past.
Developing A Growth Mindset
In our company, we saw the value of this trend when our cloud-native legal e-discovery product started to gain traction in the market. We wanted to double down on our investments in cognitive analytics to learn continuously from the market and improve the customer’s experience of our solution. One of our big challenges was providing enough holistic case-related information to litigation attorneys upfront so they could see patterns and holes in the data and find relevant or responsive case documents faster.
In the old world without serverless cloud computing, we would have needed to invest in huge hardware and on-premises machine learning tools to even start working on a data science project. But in the new world, our data science team brainstormed specific algorithms that could be used to solve various problems, such as document clustering and term frequency-inverse document frequency, a popular natural language processing concept that helps summarize documents and identify highly relevant keywords in documents.
Because of our core competency in serverless cloud computing, our software engineering, data science and product teams could generate machine learning environments very quickly. We also weren’t starting from scratch but instead using the existing base algorithms and building on them, which made our iterations faster. Our teams could do both internal and external customer experience tests with different control groups before finalizing the solution that would move on to production. This level of agility in data science innovation would be almost impossible without serverless cloud components.
Serverless cloud computing makes innovation more affordable and accessible to all companies and teams – regardless of size and resources. And with more innovation, we all benefit from the diversity of new ideas and options. The building blocks we need already exist in the serverless cloud; we don’t need to spend our precious time and resources making them. All we have to do is figure out how to use them in creative ways to benefit our companies and our customers.