5 min

Tags in this article

, , , ,

With generative Artificial Intelligence (gen-AI) rarely out of the technology newswires for a moment, current discussion around how and where it should be applied inside technology teams is rife. While gen-AI tools can indeed generate code snippets and perform certain (indeed much-needed) testing and debugging procedures, they are also argued to fall somewhere short of having the deep understanding and logic needed to write efficient and optimised code. Could the chance to re-embrace core DevOps principles and practices provide us with the effective control mechanisms needed to bring gen-AI coding forward into our cubicles and play a meaningful part in the software engineering team? One man who thinks so is Chris Astley, global head of engineering for data & AI solutions at KPMG.

While market analysts make grand predictions around how many billions (yes, billions, not millions) of Euros might be added to economies in this region as a result of generative AI, it feels like organisations themselves have not yet aligned themselves to prepare for and adopt many of these new tools. Astley and team suggest that, alongside good data management and governance, DevOps can be a critical enabler of generative AI. 

Smoothing stumbling blocks

Why is DevOps so important here? Because, he says, it removes barriers between development and operations teams, ensures rapid iteration of new models and enables organisations to adapt swiftly to changing market demands. But most are at varying stages of their journeys and there are many barriers preventing businesses from cultivating the DevOps culture they need to support their Gen AI ambitions. So, what are these stumbling blocks, and how can organisations overcome them so that their DevOps teams run as smoothly as possible?

“Time and time again, DevOps teams drift away from other departments, resulting in a disconnection where the technology being created internally does not support the needs of the organisation. By fusing DevOps teams with other business lines, it helps solve this, whether it is AI-related or not. A business model that places teams that are accountable for high-quality service outcomes close to their stakeholders is critical to success,” advised Astley, speaking to press & analysts this month.

Managing accountability changes 

But as with any reorganisation, when combining the Development and Operations teams, inevitable role changes will follow. Speaking from experiences seen across KPMG customers, partners and internal departments, Astley reminds us that a “head of build” and a “head of run” position would typically no longer exist in a traditional DevOps team, especially in the era of gen-AI. So as these jobs change, it is easy for new roles and responsibilities to blur. Clarifying accountability is vital, as without it, inefficiencies and mistakes could creep in, which may be detrimental to Gen AI development.

“The building blocks of a stellar DevOps team should focus primarily on technical workforce requirements like full stack developers, cloud engineers, data scientists, etc. However, other teams need support in areas, such as governance, cybersecurity and finance where change is required to make DevOps a success,” said Astley. “To do this, employees on DevOps teams should also have a broad set of softer skills around stakeholder management, communication, leadership, accountability, and maintaining quality, to bring the rest of the organisation along and be a point of contact for any questions or feedback about Gen AI tools.”

Preventing ‘hidden underperformance’

Astley and the KPMG team also state that while a DevOps culture often accentuates excellent performers, it can also provide a cover for underperformance because the unit of measurement is typically the team rather than the individual. This could be argued to ultimately reduce the speed to market of any gen-AI solutions being developed internally. 

“Often, underperformance can be explained and resolved, so enabling leaders to do that through proper HR processes can tackle the problem. It’s also important to be open and honest about performance objectives and use metrics to measure individual contributions, not only team-level outcomes,” said Astley. “With the gen-AI race in full swing, the scarcity of talent has never been more obvious. Being able to utilise talent pools across Europe, at scale, particularly when even further abroad, is key to success and the only way to build a high-quality practice that can support an organisation’s gen-AI goals.”

The narrative here from Astley suggests that creating a DevOps team can have [usually positive, but potentially negative is mishalndled] ramifications across an entire organisation, so it’s vital to have the C-suite on side. However, it isn’t the easiest concept to understand, even for the most seasoned executive. 

The KPMG team remind us that a good phrase [that the C-suite can understand] is “you build it, you run it”… so for the C-suite, that says the business is moving away from having two distinct parts of the technology function that handle the build and run separately, into a combined set of teams who ultimately do both. Then it’s about articulating the benefits of this model in terms of cost of delivery, agility, time to market and alignment to business outcomes, whether gen AI-related or otherwise.

Who is most at risk?

“The impact of DevOps adoption varies across sectors and organisations. While digital native companies born in the past two decades readily embrace DevOps practices, older businesses with legacy systems and structures face significant challenges in transitioning to this new model. This could potentially hold them back from benefiting from emerging technologies such as gen-AI. Additionally, those heavily reliant on third-party managed service providers will likely encounter further complexities in aligning their objectives with those of external entities,” said Astley.

He further states that cultivating a DevOps culture is paramount for organisations seeking to harness the transformative power of generative AI. “By addressing challenges, aligning with business needs and investing in the right talent and training, companies can unlock the full potential of DevOps and drive growth, innovation, and success in the digital age,” he concluded.

AI is all about the humans

Could there be some kind of AI-driven DevOps renaissance that needs to happen for us to control, corral and coalesce existing software application development practices and their operations counterparts for the greater good? It’s not an unreasonable suggestion, especially given the initial fright that many people have had with AI (the whole rise of the robots scenario) and the fact that we need to talk about keeping a human in the loop at all the right junctions and decision-making points. KPMG is certainly expansive enough to speak from experience and help raise this discussion.

So then, the true development of effective AI will come down to humans, people and teams. Who saw that coming?