BLOG | OFFICE OF THE CTO

State of Application Strategy 2024: Generative AI Redefines Automation Evolution

Lori MacVittie 缩略图
Lori MacVittie
Published May 14, 2024

In the past, the most digitally mature organizations used automation to execute scripts based on well-defined business and operational objectives to make configuration changes and push adjustments to policies. The introduction of generative AI has made that nearly passé, with the new goal now a more autonomous system.

That’s the conclusion I draw based on some recent reading of Intel’s State of Cloud Optimization 2024 in which “60% of those surveyed consider an autonomous nature in optimization tools to be very important to extremely important.”

If you’re wondering what optimization tools are, well, those are related to costs and, more specifically, reaching the top priority for 2024 of “cutting cloud costs.”

FinOps, basically.

But autonomy is not just about cutting costs in the cloud. It’s also about eliminating operational toil and leveraging the power of generative AI to autonomously execute delivery optimization and security, too.

When asked about the greatest value respondents were looking for from generative AI, the top answer for both app security and delivery uses was autonomous adjustments.

  • Security: automatically adjust security policies and generate security configurations on threats detected
  • Delivery: automatically adjust app and API policies based on service-level objectives (SLOs) for delivery optimization

In other words, enterprises are looking to generative AI to up the ante for automation strategies from automated scripts to autonomous adjustment and generation of policies to manage security, performance, and cost-control objectives.

The State of Automation

More autonomous operations may sound a bit futuristic and unachievable, but existing levels of automation across enterprises are actually fairly impressive. That’s doubly true when you consider where organizations were in their automation journey last year. 

Near the end of 2022, the majority (52.5%) of organizations were still operating under hybrid automation strategies. They used scripts to make configuration changes and push policies, but they executed those scripts manually. Only about a quarter (25.4%) were using systems to initiate scripts, and a whopping 21.9% were not using any automation at all. 

Fast forward to the end of 2023 and most organizations have made incredible progress on their automation journey.

Levels of automation chart

This year, less than one in ten (8.2%) were ignoring automation and the percentage of organizations that are achieving what was the highest maturity (automated), had nearly doubled.

But you’ll note my handy-dandy chart has a new goal: autonomous.

And that’s because generative AI showed up and pointed out that the goal of fully autonomous operations was not only possible but achievable. That’s because generative AI may hallucinate when asked open-ended questions, but when focused in on generating structured content—like configurations and code with well-formed schemas and syntax specifications—it performs well. And the more it does, the better it gets. It learns, after a fashion, and can become as proficient at generating the right policies and configurations to tune infrastructure and app services to meet defined service-level objectives.

And, based on our research, organizations want to achieve it. Likely because the benefits organizations are seeing from their automation efforts are greater when automation is more fully leveraged.

Benefits of automation chart

But all this automation must be driven by something. Something concrete. Something measurable. Something actionable.

That something is data. Telemetry, to be specific, generated by the systems and services that support the applications and APIs organizations want to ensure are fast, available, and secure.

That’s why when we talk about the six key technical capabilities organizations need to accelerate digital transformation, automation and observability go together into a single domain. Because the former without the latter is just guessing, and the latter without the former fails to take advantage of realizing the visibility organizations have desired for decades.

And all that data, all that telemetry, can also fuel the predictive AI engines that will analyze it and produce actionable insights so generative AI can use it to adjust configurations and policies autonomously.

Oh yeah. It’s all coming together now. AIOps is the next evolution of automation, and it’s all thanks to generative AI.