DEVOPSdigest

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 11 of this series, the experts provide recommendations on how to get started using AI to support software development.

START NOW

Start doing it. LLMs can be an expensive adjunct but all developers should be familiarizing themselves with what is available. Chances are many are already doing so, so look to learn from this by sharing experiences. We are at a very low level of maturity, so there is advantage to be had in accelerating experience.
Jon Collins
Analyst, Gigaom(link is external)

Your competitors are certainly looking at tools that help create code and tests as well as AI-augmented tools that make testing faster, more efficient, and less noisy. It's important to look at what's available and see how to fit it into your SDLC.
Arthur Hicken
Chief Evangelist, Parasoft(link is external)

Companies should not hesitate any longer on using AI development tooling in business. While the technology is in a boom phase thanks to LLMs, the core concepts of AI have been evolving for years. Companies that take the "wait and see approach" are likely to fall behind their competitors as others take risks today. Finding the right fit for AI in development will require trial and error. In addition, what works for one organization may not work for another. Fine-tuning AI at an early stage could form an early lead that is hard for the competition to match.
Ed Charbeneau
Developer Advocate, Principal, Progress(link is external)

Companies should actively promote the use of AI for development now, as it represents a significant shift in the pace and economics of development that businesses cannot avoid.
Peter White
SVP of Emerging Products, Automation Anywhere(link is external)

EVALUATE YOUR CURRENT PROCESS

AI has a comprehensive and exciting range of use cases, but rushing in without adequate planning can scupper a company's chances of successful AI adoption. First, evaluating your current development processes, infrastructure, and skillset is essential. Identify areas where AI can impact developers and wider business success most.
Dotan Nahum
Head of Developer-First Security, Check Point Software Technologies(link is external)

Companies looking to leverage AI in development would be wise to begin their journey by assessing their current development processes and identifying certain areas where AI can provide maximum value.
Jobin Kuruvilla
Head of the DevOps Practice, Adaptavist(link is external)

DETERMINE WHAT YOU WANT TO ACHIEVE WITH AI

The most valuable investment a company can make is determining what they want to achieve with AI. For example, do you strive to improve code quality, accelerate development cycles, or automate repetitive tasks in a certain department?
Dotan Nahum
Head of Developer-First Security, Check Point Software Technologies(link is external)

Code AI is real, but there is also a lot of hype. Pick the use case scenarios where AI can provide significant returns (based on your business priorities and tech stack) and test whether your code AI tools can provide the right kind of gains — faster time to project completions, unlocking harder transformative projects, etc. Code AI makes for cool demos, but what really matters is whether it is solving YOUR problems.
Raman Sharma
CMO, Sourcegraph(link is external)

IDENTIFY THE RIGHT USE CASES

To successfully implement AI, leaders must identify appropriate use cases where AI tools can add value to an organization.
Hao Yang
VP, Head of AI, Splunk(link is external)

Companies should start by identifying repetitive tasks that AI can automate, invest in training for AI integration, prioritize high-quality data for AI models, and maintain a flexible approach to adopting new AI tools as they evolve. This will be a process, and there will be a learning curve, but with the right approach and the right tools, AI can be a game changer for software development teams.
Anand Kulkarni
CEO, Crowdbotics(link is external)

Rather than jumping all in, choose the use cases and projects that make the most sense to prove out AI's capabilities, capacities and limitations. Based on your experience, and perhaps ROI findings, proceed with the next use case or project. Remember to consider the bigger picture in your experiments — governance, CI/CD pipeline impact, production support, security, etc. Scott Willson
Head of Product Marketing, xtype(link is external)

Select a few high-impact use cases identified by the task force for implementation as PoC or PoV that can show value in one or two quarters. Explore how vendors can help fund these activities to mitigate risk and ensure a successful PoC.
Patrick Doran
CTO, Synchronoss(link is external)

ALIGN AI INITIATIVES WITH BUSINESS GOALS

To effectively leverage AI in software development, companies should start by clearly defining their goals and aligning AI initiatives with business objectives.
Rahul Pradhan
VP of Product and Strategy, Couchbase(link is external)

DESIGNATE AN INTERNAL CHAMPION

What are you waiting for? If you haven't started already, you are behind. Find or hire an internal champion with domain knowledge and start a program to enable all employees (not just developers) with these tools.
Sakshi Garg
Head of Engineering, Hydrolix(link is external)

ATTAIN BUY-IN FROM ALL LEVELS – ESPECIALLY C-SUITE

For companies interested in investing in AI to support software development goals, it is important that teams break down organizational silos and collaborate on a path forward. There must be buy in from all levels, especially from the C-suite. Use case studies and metrics to demonstrate the tangible benefits of AI solutions, such as faster time-to-market, higher quality software, and improved customer satisfaction. Highlight how AI can further enhance these benefits and drive competitive advantage. When communicating the value of generative AI in DevOps to the C-suite, it's crucial to focus on the business outcomes and strategic advantages it can deliver.
Chetan Conikee
Co-Founder and CTO, Qwiet AI(link is external)

BE CONSISTENT

There needs to be consistency, starting at the top. If companies don't take a unified approach, each team will follow a different strategy, and the disorganization will open the door for more risk.
Udi Weinberg
Director of Product Management, Research and Development, OpenText(link is external)

FACILITATE CROSS-FUNCTIONAL COLLABORATION

Encourage cross-functional collaboration between developers, data scientists, and AI specialists to ensure the successful integration of AI into development processes.
Ed Frederici
CTO, Appfire(link is external)

Organizations should start by facilitating conversations between their technical AI teams, legal teams, and their AI service providers. Establishing a baseline within an organization and developing shared language can go a long way in deciding where to focus and minimize risk with AI.
David DeSanto
Chief Product Officer, GitLab(link is external)

Assemble a cross-functional team comprising individuals from engineering, product management, marketing, and customer support to gather use cases they would like to see these tools utilized.
Patrick Doran
CTO, Synchronoss(link is external)

EMBRACE COLLABORATION BETWEEN MAN AND MACHINE

The most successful organizations will be the ones that embrace a collaborative approach with AI, leveraging the strengths of both humans and machines to drive innovation.
Udi Weinberg
Director of Product Management, Research and Development, OpenText(link is external)

INTEGRATE INTO EXISTING WORKFLOWS

Rather than overhauling all of their software development processes at once, companies will get the most out of their AI deployments by integrating new AI tools and use cases directly into their existing workflows. By keeping the interfaces, systems, and workflows that employees are already working with the same, there will be less disruption in developer activities, thus decreasing the adoption curve by developers.
Marco Santos
Co-CEO, GFT Technologies(link is external)

START SMALL

Start small and experiment using AI codepilots as assistants to application authors. As these models improve, experiment with different tasks and evaluate the results. Look for the low-bearing fruits, such as repetitive tasks or tasks that are time-consuming but simple.
Karl Cardenas
Director, Docs & Education, Spectro Cloud(link is external)

Begin with specific, well-defined use cases where AI can add immediate value, such as automating testing, code review, or bug detection. As the team gains experience and you are yielding the right results, gradually expand.
Ed Frederici
CTO, Appfire(link is external)

START WITH A PILOT PROJECT

Pilot projects can be an effective strategy to help employees gain experience with AI tools and see positive results, which also translate favorably for senior leadership teams deciding whether to make AI a permanent staple. You can use these pilots as an opportunity to gather constructive feedback from developers, ensuring that the AI tools you choose align with your business goals and integrate well with your existing technology stack. For example, you might choose tools that rank highly among developers for ease of use, scalability, customization options, and support.
Dotan Nahum
Head of Developer-First Security, Check Point Software Technologies(link is external)

BUILD CENTERS OF EXCELLENCE

I encourage developers to build centers of excellence within their companies that can help solidify and share out best practices for when and how to use AI effectively.
Jeff Hollan
Head of Applications and Developer Platform, Snowflake(link is external)

CREATE MICRO-COMPANIES IN THE ORGANIZATION

AI-generated code and apps will allow startups to quickly catch up and even surpass established vendors in less time than has ever been previously possible. Given this threat, it would be wise to cultivate entrepreneurial-minded employees by creating incubation projects — micro-companies within your organization that monetarily reward these "companies" for producing profitable innovations. To compete as an enterprise today, you need to democratize innovation. A traditional top-down approach is too myopic and leaves enterprises vulnerable to being blindsided and replaced. With AI, no company is too big to fail.
Scott Willson
Head of Product Marketing, xtype(link is external)

START WITH LOW-RISK TASKS

I recommend starting with low-risk tasks — human oversight is still essential for large-scale, high-stakes projects like major refactoring or production migrations and probably will be for the foreseeable future. But being able to bootstrap a v0 of that dashboard, admin interface, or helper tool you just haven't entirely found the time to build is a valuable application of AI.
Michael Webster
Principal Software Engineer, CircleCI(link is external)

My advice is not to try to build Rome in a day. Integrate some basic AI code suggestion tooling or AI code testing, in areas where they won't derail current processes if you encounter issues. If they provide value, expand from there.
Shomron Jacob
Head of Applied Machine Learning & Platform, Iterate.ai(link is external)

LOOK TO SUCCESSFUL EXAMPLES

Be sure to absorb advice from others in the industry going through this same transformation. For instance, companies like Microsoft and Google have successfully integrated AI into their development processes, and offer useful case studies on their approaches.
Shomron Jacob
Head of Applied Machine Learning & Platform, Iterate.ai(link is external)

Go to: Exploring the Power of AI in Software Development - Part 12: More Recommendations