Software teams can now turn ideas into functioning applications faster than ever. AI coding assistants are contributing to that acceleration, but a working application is not necessarily one that is secure, resilient or ready for the demands of production. The challenge for businesses is to strengthen the engineering practices around software as the pace of change increases.
A study of 4 867 developers at Microsoft, Accenture and an anonymous Fortune 100 company found that access to an AI coding assistant increased completed tasks by approximately 26% across three field experiments. It is evidence of a real productivity opportunity, although faster task completion does not, by itself, establish software quality.
Google Cloud’s 2025 DevOps Research and Assessment (DORA) research likewise reported a positive relationship between AI adoption and software delivery throughput, but a negative relationship with delivery stability. The research points to the importance of automated testing, version control and fast feedback loops when teams increase the volume of changes they make.
For South African cloud platform engineering company KineticSkunk, the question is increasingly what happens after code is produced: who validates it, who operates the supporting cloud platform and who remains responsible as the application changes?
Now that everyone’s a coder, who’s checking the code?
Donovan Mulder, founder and CEO of KineticSkunk, is enthusiastic about what agentic AI makes possible. The company is exploring how Amazon Web Services (AWS) agentic technologies could accelerate elements of its engineering services. But Mulder is equally focused on the quality of the software businesses are producing, regardless of whether the code was written by an experienced engineer or generated with AI assistance.
A prospective customer operating in a regulated environment had developed an application using AI coding tools without conventional software engineering expertise. The application functioned, but that alone did not answer questions about its security, resilience or readiness for production. The situation illustrated the need for an independent engineering assessment before proceeding.
Such an assessment can involve code review and infrastructure evaluation to understand how an application is built and protected, as well as penetration testing to examine weaknesses from an attacker’s perspective. These are complementary forms of assurance, not evidence that a functioning application is safe simply because it passes a demonstration.
we need to recognise the difference between trained engineers using AI to work faster and inexperienced software developers using AI to bite off more than they can chew.
KineticSkunk CEO Donovan Mulder
“You don’t have to stop building with AI,” says Mulder. “I think it’s one of the most exciting developments in software engineering. But we need to recognise the difference between trained engineers using AI to work faster and inexperienced software developers using AI to bite off more than they can chew. They don’t necessarily bring the same appreciation of security or what happens when something goes wrong.”
The concern is not limited to newcomers. Stack Overflow’s 2025 Developer Survey found that 46% of respondents distrusted the accuracy of AI-tool output, compared with 33% who trusted it. AI can help experienced engineers too, but experience does not remove the need for review, testing and accountability.
“An application can function perfectly on the surface,” says Mulder, “but it might carry vulnerabilities underneath. If you build proper verification into the development process, you can keep innovating but still have security and quality that can keep up with that pace.”
When the project ends, the responsibility begins
KineticSkunk has built its cloud engineering expertise through focused professional services across AWS and Azure, including migration, modernisation, security and resilience. Those engagements continue to solve immediate customer problems. What is evolving is the relationship between the initial engineering project and the responsibility for a platform once it is in use.
A platform assessed as secure and resilient at handover may behave differently six months later. Teams add features, change infrastructure, introduce dependencies and alter access. Google’s 2025 DORA research describes AI as an amplifier of existing organisational strengths and weaknesses. Development can accelerate, but the systems and practices supporting delivery must mature alongside it.
KineticSkunk’s flagship Managed Platform Solutions proposition addresses that longer-term need through ongoing responsibility for an agreed platform scope. Depending on the engagement, that may include monitoring, incident response, controlled releases, security and configuration reviews, recovery readiness and continuous improvement. The company works alongside internal engineering teams or assumes clearly defined operational responsibilities. Its AWS managed offering brings those disciplines into the cloud environment that supports the application, rather than treating the project handover as the end of the conversation.
Beyond keeping the lights on
KineticSkunk’s “Store. Protect. Prove.” approach offers an example of why ongoing assurance matters.
“There are backups, and then there’s recovery,” Mulder explains. “You can back up your data religiously, but that doesn’t mean you can recover the application. And when your developers are continuously introducing new features, you’re changing what needs to be recovered. For me, resilience is something you have to prove. That’s why resilience needs to be part of an ongoing operating relationship. Far too often it’s just something you revisit when there’s a crisis.”
The same principle applies to software security and quality. As releases introduce new code and dependencies, recurring code analysis, automated and functional testing, security testing and controlled remediation help teams find problems before they become operational incidents. Consequential changes still require appropriate human review and approval.
This is where KineticSkunk’s Platform Assurance capabilities fit. Quality engineering tests how software behaves; security testing examines exposure; observability helps teams understand what is happening in production; and resilience testing checks how systems respond to disruption. Together, these disciplines support an ongoing managed operating model without suggesting that every customer needs every capability.
KineticSkunk has explored similar questions in its AWS Insights series, including cloud operations, AI agent security and the business value of well-architected AWS environments. The common thread is the connection between engineering decisions, ongoing responsibility and outcomes that matter to the business.
The future belongs to those who can keep up
AI is changing the economics of software creation. More people and teams can develop useful applications and bring ideas to market quickly. But the ability to build faster makes it more important to validate changes, understand dependencies and keep operational safeguards in step.
Mulder sees the next generation of managed cloud partnerships emerging around this challenge: combining the speed of modern engineering with the discipline required to operate business-critical systems.
For KineticSkunk, the real measure of technological progress may no longer be how quickly we can build something, but how confidently we can keep building on it.
