Report: The Impact of AI on DevOps Workflows

Industry Survey & Insights From technology Leaders

Key Findings

  • 1 in 4 companies expect deployment frequency to increase by 90% or more as their teams incorporate more AI tools in areas like testing and automation.
  • Small and Mid-Size tech firms are 56% more likely to have already adopted AI and GPT tools in DevOps workflows in 2023.
  • Use of AI tools correlates with company size. 84% of SMBs reported AI use in DevOps, while almost half of Enterprise companies have not yet incorporated AI or GPT tools.
  • 72% of tech companies surveyed anticipate AI tools will have a high impact on deployment frequency in Q2–Q4 of 2023.
  • Only 2.5% of companies reported outsourcing DevOps fully, suggesting staffing impacts of AI tools will mostly be felt by in-house teams.

Summary of Findings

The purpose of this report is to explore how companies are adapting to the rapid growth of DevOps through the use of AI tools.

The data presented in this report is based on survey input from 120 leading tech firms including Globant, Red Hat, and Veeam. With a market size estimated at $10.4B and growing rapidly towards an estimated $25.5B in 2028, DevOps is a key area for staffing growth in the tech industry; as well as a key area for the use of AI and GPT tools.

Most notable in our findings was the degree to which tech companies believe AI and GPT tools will increase their DevOps outputs (deployment rates) in 2023–2024; 25% of surveyed companies indicated an expectation that deployment rates would nearly double. Meanwhile, only 12% expected "low or no impact" within a one-year timeframe.

Also notable was the implementation of AI tools in existing workflows when analyzed by company size. 71% of companies indicated that AI and GPT tools were already a major part of their DevOps workflow, but Startups and Mid-Size tech companies displayed higher adoption rates than Enterprise-scale companies. We expect this gap will close and follow similar technology adoption patterns with larger companies implementing new tools once the risk profile has been proven out by smaller and less risk-averse firms.

The results below visualize our findings in full, including breakdowns by company size and category when appropriate.

Zoom In: AI Impact on DevOps Teams

EdTech Companies by Revenue

DevOps teams surveyed reported primarily fully in-house teams, with only 2.5% reporting that the function was fully outsourced. More than half (57.5%) of companies used fully internal teams, while the second-largest segment (40%) used a mix of in-house and nearshore or offshore resources.

Traffic, Engagement, Sign-Up and Revenue by Revenue Size of EdTech & Online Learning Companies

Companies surveyed leaned heavily into automation and testing when describing their current and planned use cases for AI. Core task areas with financial or legal repercussions for errors like cybersecurity and project management saw the lowest level of survey results; compared with Enterprise, small companies and startups were twice as likely to list project management and cybersecurity as current or planned application areas for AI.

Revenue Size for Traffic, Engagement, Sign-Up and Revenue

Deployment frequency was described by participants as the key metric AI is expected to impact; managers surveyed described AI as an efficiency tool for staff as the DevOps market size increases, not as a staff replacement. Impact expectations were consistent across company stages and scales. The majority of surveyed companies (71.67 %) indicated moderate to high anticipated impact of AI tools on their deployment rate.

Current AI Adoption Rate Differences Between Startups, Mid-Size, and Enterprise Companies

EdTech & Online Learning Companies by Employee Count

Among the 72% of companies that reported AI tools as a major compentent of their existing DevOps workflows. Startups were 56% more likely than Enterprise companies to indicate current usage.

Specifically, we saw 84% of SMB and Startups indicating active adoption of AI tools in production, compared with 70% of Mid-Size companies and about half (54%) of Enterprise-scale companies.

This finding is consistent with the risk profile of AI tools, where we see smaller, newer, and less entrenched companies such as OpenAI leading the charge into GPT tooling while incumbent players like Google are more cautious about public releases.

AI Impact on DevOps: Leader Insights

Sarwar Raza, VP & GM, Cloud Solutions, Red Hat

A critical way that AI will impact DevOps is the improved ability to integrate machine learning models directly into DevOps workflows and treat models like code artifacts.  MLOps and DevOps will blend together, beginning with the evolution of AI platforms and tooling that span the hybrid cloud, helping to break down silos between data scientists and intelligent application developers.

On the AIOps side, organizations, aided by technology vendors, will improve their ability to apply AI/ML to DevOps to improve code quality, and quickly identify issues before things go into production.  Discovering these leading indicators through the use of AI will speed up application deployment, reduce risks and help control costs throughout the entire DevOps pipeline.

Wing To, VP of Engineering for Value Stream Delivery Platform & DevOps,

AI is the key to unlocking the value of DevOps. The promise of DevOps  –  faster delivery of higher quality software that is more aligned with the needs of the user – has driven mass adoption across organizations, and many have now reached the point of automating software releases and deployment into production.  

There is now a shift to understanding and optimizing the value of this transformation – optimizing the effectiveness of delivering the software, and optimizing the value of the software being delivered.  The standard approach is to gather metrics from across the DevOps pipeline: for example, time to value, deployment frequency, flow times to name a few.  However, organizations face a challenge with what to do with all this data. Aggregating the data into high-level summaries makes the information unactionable, looking at the data across 1000s of releases is too much to consume.  

This is where AI can step in and identify trends and patterns that are actionable for a particular application or teams across the vast lakes of data that is available, as most DevOps processes are automated. An example of how AI can improve quality includes predicting the rate of risk of putting a change into production. Similarly, AI can drill into the data to identify areas of bottleneck or predict risks to deliveries.

Dr. Chris Mattmann, Chief Technology and Innovation Officer, NASA's Jet Propulsion Laboratory

The biggest change that I see is that AI has the potential to become that CI/CD (continuous integration/continuous delivery) bot that we’ve all dreamed of in software for the last decade. Imagine Jenkins and/or TravisCI, but on steroids, recognizing when a build has failed, analyzing why, adapting some script (on its own), and then redeploying tests, integration, and so on. AI has the potential here to be a game changer.

Dave Russell, VP of Enterprise Strategy, Veeam

When it comes to AI, automation is likely to drive significant change within the DevOps space in the next five years.. For organizations that support a large number of customers with very diverse environments, testing all possible paths and deployment + infrastructure & cloud combinations are literally impossible. The automated testing for proper application expectations is a benefit, but when applied to security and corner-case verification, this can help keep the speed of delivery up, while better containing the risks.

While the potential impact AI may have on the DevOps space is very high, I’ve learned over the decades that broad usage and deployment always takes longer than anticipated. Also, AI/ML is a lot like a calculator. The point is to reduce errors, and increase speed, but not necessarily remove all knowledge and responsibility from the human. Skilled humans will still be very much needed; however, empowering more of the IT staff to achieve more, accelerate, and to deliver safer solutions is why I anticipate high impact.

Nicolás Ávila, Chief Technology Officer for North America, Globant

The DevOps space will experience several significant changes over the next five years due to the progression of AI.

Some key things that we can expect to see include:

--> AI-driven tools will help accelerate the adoption of a shift-left testing approach, as well as make continuous testing and delivery a much easier and faster process.

--> The developer and tester experience will be enhanced by AI as we’ve recently seen with GitHub’s unveiling of CoPilot X and Globant's own Augoor. By analyzing builds, test results, and information from different tools integrated in the pipeline, AI tools will be able to provide smart recommendations, like code refactoring opportunities or new tests to be added.

--> We’ll also see a wider support range for programming languages and frameworks. One of the challenges of working with DevOps tools is knowing a variety of different tools and frameworks, each of which is constantly being improved and extended. AI tools will accelerate the fulfillment of DevOps’ original promise: to enable the developer to bridge the gap to operations. It will be the shift-left of the development community, from testing to system configuration to improving the security of those systems.

--> Other promising areas are predictive maintenance and production failure prevention. An enormous amount of data is being generated by an ever-growing number of interconnected systems. When something fails, there’s a domino effect that results in an alert flood, making it difficult to separate the signal from the noise. AI will help DevOps teams pinpoint the root cause of the incidents faster, which translates into quicker resolution times.

Abhi Shrikhande, VP and GM of Technology Services, Toptal

DevOps is a set of practices that promote continuous improvement and collaboration across the software development lifecycle through automation. Advances in AI and Machine Learning will continue to improve automation, increase efficiency, and reduce human errors in provisioning and configuration, software deployment, testing, infrastructure monitoring, etc.

AI tools might be able to recommend improvements and optimizations to DevOps processes, particularly in identifying and leveraging an organization's best practices. The power of AI will also come to bear in creating self-healing systems that use pattern recognition and analyses of historical data to predict and resolve potential issues before they occur. As AI capabilities for managing these underlying infrastructural needs evolve, it will enable people to spend more time on innovation and value-added activities.