A Guide to Generative AI for DevOps Teams

DevOps managers should understand that integrating AI into DevOps is a major paradigm shift and that it takes time to bear all of its fruit.

There has been copious talk over the past couple of years about how DevOps teams and practitioners can benefit from generative AI. The ecosystem has been abuzz with discussions of leveraging GenAI to accelerate virtually all aspects of day-to-day DevOps work, from coding to software testing to documentation generation and beyond.

A topic that has received much less attention is how DevOps team managers – as opposed to practitioners – can use generative AI. That’s a challenge because to leverage GenAI at scale, businesses must do more than cross their fingers that individual DevOps engineers will figure out how to use AI effectively. Instead, they need DevOps leadership to oversee a systematic approach to implementing and optimizing AI adoption by their DevOps teams.

As someone who has helped DevOps teams make the pivot to a GenAI-centric strategy, I’m here to offer thoughts on how DevOps leads can help guide the process. Every team’s GenAI challenges and opportunities are different, of course, and there is no one-size-fits-all approach to ensuring that DevOps teams can take full advantage of GenAI technology. But there are several best practices for DevOps managers to consider following.

Why and how can DevOps teams use generative AI?

Before delving into specific ways that DevOps team managers can help steer engineers toward effective adoption of GenAI, let’s discuss what GenAI can do for today’s DevOps practitioners.

The short answer is that generative AI has emerged as a major productivity booster – you might even call it a pair programmer – for DevOps engineers. To date, no AI tool has emerged that is reliable and flexible enough to supplant DevOps workers fully, but AI can add substantial efficiency to the way DevOps engineers work.

For example, by using AI to generate boilerplate code, DevOps engineers avoid having to copy and paste from code templates and then tweak the results to fit a given use case. Likewise, they can ask AI to explain how code works, which accelerates the process of interpreting code from third-party sources. They can even benefit from GenAI for tasks like generating test cases, which they can, in turn, use to test their code – whether it’s AI-generated or written the old-fashioned way – and ensure it works as expected.

Competent DevOps engineers can complete tasks like this without AI, of course. But AI makes them faster, while also reducing the cognitive load incurred by practitioners.

See also: AI Workloads Need Purpose-built Infrastructure

Best practices for managing GenAI adoption by DevOps teams

In most cases, leveraging GenAI for DevOps use cases like these doesn’t happen automatically or organically. Engineers need deliberate guidance from DevOps leadership on how to approach AI, as well as how to integrate it efficiently and securely into their workflows.

The following practices can help DevOps team managers deliver that guidance.

1.    Roll out AI capabilities

It may seem obvious that a DevOps team can’t adopt AI very well if AI tools are not available to it. But too often, I’ve encountered environments where businesses place strict limitations on which AI technologies employees can use, typically in a bid to mitigate security and data privacy risks associated with AI (like the potential for proprietary code to leak to third parties when engineers don’t use AI tools securely).

These risks are real, but they are not a reason to avoid using AI at all. Companies that take that approach shoot themselves in the feet and will end up missing out on the massive DevOps productivity gains that come with AI. As a result, organizations will lose competitive advantage to firms that are operating with higher productivity as a result of AI.

Instead, DevOps managers should work with other stakeholders in the business to roll out appropriate GenAI technology for DevOps practitioners – such as AI-assisted coding and testing tools. With a proper technology adoption process in place, the business can identify AI solutions that meet its needs, while also keeping security risks in check.

2.    Allocate sufficient budget

Along similar lines, AI tools for DevOps often cost money. Management must be willing to allocate sufficient financial resources to support GenAI tool adoption.

If a push to invest in GenAI solutions leads to blowback from other stakeholders in the business, DevOps managers should stress the productivity boost that is likely to stem from AI and encourage other leaders to view AI as a means of saving money in the long run.

3.    Adopt GenAI tools for management tasks

In addition to overseeing the adoption of GenAI by DevOps engineers, team managers can leverage GenAI themselves for tasks such as drafting emails, creating presentations, and summarizing content.

The more hands-on experience managers gain with real-world use of GenAI, the better positioned they’ll be to help guide staff working under them in applying AI to their roles.

4.    Provide oversight (especially for junior engineers)

DevOps engineers – especially less experienced junior engineers – need oversight to do their jobs effectively in general. This becomes all the truer when they begin using AI.

Typically, practitioners won’t know at first how to guide GenAI tools appropriately to achieve desired results. They may not know how best to verbalize the instructions they feed to AI to generate or explain code, for example. They may also struggle to recognize errors in AI-produced code.

A primary responsibility of DevOps managers overseeing a pivot to GenAI is to supervise how practitioners use AI tools and identify challenges like these. Simply expecting engineers – especially junior engineers with limited experience – to pick up a complex new tool like AI on their own and learn to use it optimally is not realistic.

5.    Measure success in terms of ends, not means

A good DevOps manager doesn’t fixate on exactly how engineers use AI – or, for that matter, any type of productivity-enhancement technology. Instead, they focus on what the results are.

To that end, managers should assess the success of their practitioners in terms of how efficiently they can complete tasks like writing, testing, and deploying code, not how extensively they rely on AI to do these things. Some engineers are likely to find greater value in AI than others. The role of the manager is to encourage practitioners for productivity’s sake, not to achieve some arbitrary level of AI adoption.

6.    Be patient in assessing AI’s benefits for DevOps teams

Finally, DevOps managers should understand that integrating AI into DevOps is a major paradigm shift and that it takes time to bear all of its fruit. Expecting major productivity gains in just a few months is unrealistic for developers with an average skill level (although some may learn faster); a more reasonable timeline is half a year, at a minimum, before engineers become sufficiently adept at GenAI usage to demonstrate substantial improvements.

The upside is that, once AI does become an integral part of DevOps workflows, its benefits are likely to continue to accrue as practitioners learn how to take ever-greater advantage of GenAI solutions. But the initial learning curve is steep, and it’s important to set realistic expectations.

Leave a Reply

Your email address will not be published. Required fields are marked *