DEVOPSdigest

The Holiday Season means it is time for DEVOPSdigest's annual list of DevOps predictions. We are kicking off the predictions series this year with the final installments of a series run on DEVOPSdigest last month: Exploring the Power of AI in Software Development. As part of that series, DEVOPSdigest invited experts across the industry — consultants, analysts and vendors — to comment on how AI can support the software development life cycle (SDLC). In Part 14, experts offer some predictions about how AI's support for software development will evolve in 2025 and beyond.

MASSIVE EXPANSION OF AI IN DEVELOPMENT

Current trends suggest that AI's role in software development will continue to expand significantly. We can anticipate AI becoming even more sophisticated in understanding and generating code, potentially automating more complex tasks that currently require human intervention.
Ramprakash Ramamoorthy
Director of AI Research, ManageEngine

The use of AI in software development will expand. Nearly 40% of respondents in our 2024 Global DevSecOps Report said they currently use AI in software development, but double that (78%) expect to use or plan to use AI in the next two years.
David DeSanto
Chief Product Officer, GitLab

We are in the early stages of AI usage in software development. The usage will expand 10x in the next 5 years.
Raman Sharma
CMO, Sourcegraph

AI BECOMES UNIVERSAL

The use of AI in development will become universal. The quality of generated code will improve, though maybe not as much as we'd like, especially where security is concerned. We'll eventually get to the point where AI will be able to block out designs with UML, C4 and such.
Mike Loukides
VP of Emerging Tech Content, O'Reilly Media

I predict that every development shop will be leveraging AI in at least one area of the SDLC by 2025, be it in developer assistance, enablement, test generation, documentation generation, or product management. The opportunity to accelerate innovation and increase developer productivity is massive and the tools continue to get more secure, integrated, and powerful.
Matt Healy
Director of Product Marketing, Intelligent Automation, Pega

AI EMBEDDED ACROSS ENTIRE SDLC

AI will become more deeply embedded in IDEs (Integrated Development Environments) and other development tools, providing seamless assistance throughout the entire development life cycle. This functionality includes real-time code suggestions, automated testing, and intelligent debugging assistance.
Dotan Nahum
Head of Developer-First Security, Check Point Software Technologies

In the near future, AI will become deeply integrated across the entire software development life cycle, enhancing everything from coding and testing to deployment and maintenance.
Rahul Pradhan
VP of Product and Strategy, Couchbase

Today, many still view AI in development simply as code generation, but the transformational opportunity extends far beyond. To realize its full potential, AI must be embedded across the entire software development life cycle, allowing everyone involved in delivering secure software, not just developers, to benefit from the efficiency boost. That shift is beginning to happen now and will accelerate quickly.
David DeSanto
Chief Product Officer, GitLab

AI will be integrated into every phase of the software development life cycle. From refining project specifications, to generating coding to testing and optimizing deployment processes, AI will continue to enhance development productivity and code quality. In addition to the automation of routine tasks, AI will help strengthen security by enforcing coding best practices, generating comprehensive test cases, and monitoring application performance and security real-time. This will not only predict and detect potential issues, but automatically deploy fixes.
David Brault
Product Marketing Manager, Mendix

EXPANSION AND CONTRACTION PHASES

The use of AI in development will soon start going through phases of expansion and contraction. During the expansion phase, organizations will adopt AI tooling and quickly learn the strengths and weaknesses of LLMs. Once organizations understand the risks and weaknesses, they will contract and shift towards fine-tuning (LLMs using customized data sets). This process will resemble what happened with cloud and on-premises software hosting. The result will be AI models which use federated learning and hybrid inference, balancing local and cloud infrastructure.
Ed Charbeneau
Developer Advocate, Principal, Progress

AI TOOLS MATURE

The AI development tools we have today are the most basic, rudimentary, and worst versions we'll have in the coming decade. To draw a parallel, think of IDE tools from several decades ago. Sure, they offered a few autocomplete features and codebase management tools, but that was pretty much it. Today, IDEs have comprehensive toolkits ranging from productivity features to package management across nearly every major language and toolset you use as a developer.

AI developer tools are today where IDEs were a few decades ago. They offer some tangible benefits in a few use cases, are sometimes impressive, and have shown what a change in what it means to write software might look like. But there are so many more things that they will improve upon, such as natural language recognition, feature planning, pulling in context from your whole codebase, connecting to your deployment and source control environments, and more. All the while, models will become more powerful (and economical) over time, leading to suggestions these AI tools become ever more sophisticated and correct as well.
Phillip Carter
Principal Product Manager, Honeycomb

AI GUIDES OVERALL DESIGN

AI will develop a deeper understanding of the software architecture and design principles that guide large software projects so that its contributions align with the overall design philosophy, and perhaps eventually guide it. While many concepts and tradeoffs are held in the minds of software architects, we can expect AI to expand its abilities in this area.
Dr. William Bain
CEO, ScaleOut Software

AI HYPE COOLS

The market for AI in development is alive and awash in energy, enthusiasm, and opportunity. Of course, some of this unbridled growth and enthusiasm will cool, and AI will settle in among the other available tools within a developer and company's repertoire, as one of many other tools and toolsets that companies leverage.
Cassius Rhue
VP, Customer Experience, SIOS Technology

AI SPRAWL WILL EASE

AI crowd-out and sprawl will ease during the course of the year, enabling teams and organizations to refocus on game-changing organizational improvement frameworks like value stream management to deliver tangible business outcomes.
Helen Beal
CEO and Chair, Value Stream Management Consortium

REAL SOLUTIONS EMERGE

AI will continue to see its promise outstripped by its reality, but the real solutions — the ones that genuinely create productivity boosts and make developers' lives easier — will start to emerge and get better integrated into the SDLC.
Marcus Merrell
Principal Test Strategist, Sauce Labs

AGENTIC APPROACHES POWERED BY AI

In 2025, DevOps will undergo a transformative shift with "Agentic" approaches powered by AI. Platform-centric strategies will enhance collaboration, automate repetitive tasks, and make software delivery faster and more reliable. Agentic approaches can automate decision-making, predict bottlenecks, and enhance collaboration. They will unlock a new era of software development and delivery, giving organizations a competitive edge in the digital age.
Vishnu Vasudevan
Head of DevOps, Opsera

AGENTIC CONTINUAL ROBOTS

In more traditional software development, we'll see agentic continual robots that actively improve the code, get external information to patch security and resolve production scaling issues.
Tiago Cardoso
Principal Product Manager, Hyland

PRIVATE VS PUBLIC LLM

AI's role in application development, performance, and end user experiences will be transformed by developer teams' ability to tap into their own private LLMs in 2025. Whereas using public LLMs puts organizations at risk of exposing sensitive data — and can lead to more generalized results thanks to their generalized input — private LLMs offer teams much more control over their data.
Shomron Jacob
Head of Applied Machine Learning & Platform, Iterate.ai

PRIORITIZING ROI

As we approach 2025, prioritizing the return on investment (ROI) from AI technology implementations will be crucial for today's businesses. We know the potential of AI, but organizations must build a compelling business case for senior leaders to understand the impact around productivity improvements, higher quality outputs, overall efficiencies, etc. While it is clear that organizations will likely look to invest more in AI technology, they need to determine in which part of their businesses the technology will make the largest impact. In terms of development, for example, AI can be impactful in assisting those with minimal technical skills to generate code or can help developers achieve increased productivity so they can work on more challenging development projects.
Amitha Pulijala
VP of Product, Vonage

QUANTIFYING AI ROI

It will become easier to quantify the ROI of AI-based software development tools. Organizations have successfully demonstrated the rapid adoption of AI, but have struggled to measure its impact across diverse teams and business functions. This is, in part, due to misaligned focus. Instead of asking, "How is AI helping?" leaders should focus on specific tasks, such as test generation, documentation, or language translation, and measure the gains in efficiency and productivity for these AI-driven activities. Companies can more effectively quantify the ROI and justify further investment in these technologies by focusing on the tasks where AI excels.
Brian Wald
Global Head, Field CTO, GitLab

SMALL TEAMS CAN ACHIEVE BIG THINGS

The fundamental equation of what it takes to build and scale a software company is changing with AI-native software development and testing. Founders will no longer need substantial investment and an army of developers to build the next billion-dollar company. AI code generation and testing allow small teams to achieve big things.
Maryam Ahmed Hassani
Co-Founder and CEO, Zealous

Go to: Exploring the Power of AI in Software Development - Part 15: 2025 Predictions and Beyond