As we reflect upon the last year, it’s clear that artificial intelligence (AI) and machine learning (ML) were among the top “buzzwords” for techies everywhere. While very popular concepts, I’d argue that 2018 wasn’t exactly the year of AI and ML; it was just the beginning of a long, gradual transition – especially for DevOps.
With the number of technologies and service providers proliferating within the market (just look at how many new laptops, tablets, phones, watches, etc. that Apple, for example, released in the last year!), dev teams today are tasked with the nearly impossible demand to develop, release and update software at record speeds. Unfortunately, the traditional dev tools we normally use aren’t doing enough to help us out. They’re not automated enough or intelligence enough to help do the heavy lifting of testing, analyzing and repairing code to help us meet our dwindling deadlines.
That’s where AI and ML step in. After struggling with how to best utilize the technologies in 2018, more teams around the globe are looking for easier ways to incorporate them into their pipeline tool chain, helping to maximize productivity, deliver powerful data analytics and more in the coming year.
Below are three tips to help dev teams address rapidly accelerating release cycles with these hot new technologies.
Look at AI and ML differently.
For the last few years, I’ve found that there’s been this sort of hesitancy towards new technologies like AI and ML that is holding people back from introducing them into their current tool stack. Preoccupied with various fears of world takeovers (e.g. The Matrix) and layoffs at work, many DevOps professionals have misunderstood the power of these tools – and how they can positively impact their work.
While these solutions are often powerful enough to transform day-to-day responsibilities, it’s for all the right reasons. At this stage, AI is nowhere near a place where it can totally diminish one person’s job – never mind an entire department’s. However, it can make it easier. For example, today’s capabilities can replace monotonous, dull DevOps jobs, such as analyzing test results, so that teams can spend more time acting on those results. While the tools can analyze data and identify patterns at record speeds, they lack the human skills and experiences to determine whether those patterns are relevant.
Start small (and then expand).
As with any new technical deployment, be sure to start small (and slow) when implementing AI and ML integration. The greatest successes, in my opinion, have always started in pieces. For example, look to mobile app testing first; once successful results are achieved (and teams become more comfortable with the new tools and processes), then consider using AI and ML for web testing and IoT after that.
Use it for the right stuff.
As mentioned earlier, AI and ML are still in the very early stages. Similar to how we were looking at cloud adoption a decade or so ago, organizations today are in the process of trying to determine if, when and how these technologies can actually work within their businesses. As we learned with the cloud, it doesn’t make sense to go all-in from the get-go. Instead, there are certain day-one use cases that just make sense (and others that don’t).
When getting started, identify the complex tasks that require the littlest thinking and human reaction and use AI and ML to automate these particularly uneventful and repetitive (yet, important) duties, such as data analysis. Dev teams should also define the working processes that require modifications accordingly, such as using these technologies to drive Go/No-go decisions.
As we kick off the new year, it’s important for dev teams to approach AI and ML with a fresh outlook. While these tools are poised to completely transform the development process, it’s important to remember that it’s all about how you use them – not how much. When implemented correctly by DevOps professionals, the technologies have the ability to do more than just test automation and data analysis. In the next few months, expect to see use cases expand to include tasks such as security and code scanning, as well as code coverage. But don’t forget: To really see this come to life, it’s important that dev teams don’t blindly dive in. As we start 2019, sit down with your colleagues and figure about the best strategy to introduce AI and ML to your organization and watch magical things happen.
