Tiago Azevedo is CIO at OutSystems, a leading AI-powered low-code development platform.getty​The accessibility and affordability of AI development tools have brought us to an inflection point. Tired of the build-versus-buy binary, enterprises are increasingly opting for an alternative: AI-driven coding—or, as many in the industry call it with mixed feelings, vibe coding.​How We Got HereFor a decade, low-code was the user-friendly successor to traditional coding. It faced resistance from parts of the development community, but its staying power spoke for itself.​AI development builds on the same idea—higher abstraction, faster delivery—and pushes it further. Developers describe what they want; an AI assistant translates that into working code. The code is still there, still written in the same programming languages and still available when something needs to be refined. The difference is how much of the manual work the developer is doing directly.​The economics of custom software have changed as a result. Projects that used to require deep pockets and an extensive bench of senior engineers are now viable with a fraction of the time and resources. That changes the build-vs-buy calculus inside large enterprises, not just the build side of it.​The impact is real. IT teams can deliver on business needs faster and at a bigger scale, whether that means modernizing legacy infrastructure, standing up net-new software or pursuing projects that previously wouldn't have cleared the cost-benefit bar. The question has shifted from whether to use these tools to how to organize a technology function around them.Succeeding In An AI Development EnvironmentAs vibe coding took off, so did development projects from non-technical users. Results varied. Most first-timers landed on the same conclusion: knowing the ins and outs of an application—what it should do, how it fits the rest of the infrastructure, where it can fail—is harder than it looks from the outside. Developers are still best positioned to produce enterprise-grade software, even when they're not writing the code line by line.​A recent OutSystems report found that two-thirds of enterprises are already using AI development tools to build or extend business applications. But adoption doesn’t always equate to value. For enterprises exploring AI development, four things matter more than they used to, and three of them have almost nothing to do with code generation itself:​1. Business JudgmentWhen the cost of generating an application drops sharply, the constraint moves from engineering capacity to product judgment. The scarce skill is no longer who can build it. It's who knows which applications are worth building and how they fit the organization's actual problems.2. Infrastructure And OperationsThis is where most takes on AI development fall apart. Faster code generation doesn't eliminate the servers, the hosting, the cloud spend, the environments, the CI/CD pipelines, the observability, the secrets management or the release process. It adds pressure to all of them because the volume of software to operate goes up. Any organization serious about AI development has to be equally serious about the platform and operations function underneath it. That function doesn't shrink. It reshapes, and in most cases, it grows.3. Determinism In AI Code GenerationLow-code platforms solved this a decade ago: The same inputs produce the same outputs, in a form that can be governed, audited, upgraded and operated at scale. AI code generation is not there yet. Two prompts can produce two working-but-incompatible implementations of the same requirement.For prototypes and internal tooling, that's acceptable. For mission-critical enterprise systems, it isn't, not without the guardrails, platforms and review processes to compensate. Pretending this problem doesn't exist is how you end up with an estate of applications nobody can confidently maintain.4. Shape Of The OrganizationTaken together, these shifts don't just change the developer's day. They change the technology function: skill mix, team topology, the balance between builders and operators, vendor relationships and how headcount is allocated across engineering, product and platform. The organizations that get real value from these tools will be the ones willing to reshape around them, not just adopt them.​What's NextIn 2026, software leaders have more tools at their disposal than at any prior point in the industry. AI development, used with the right guardrails and platform discipline, is how teams keep pace with—and in some cases set the pace of—their industry. It isn't a rejection of traditional code. It's the next evolution of it, and of the developer role along with it.​The teams that win won't be the ones with the best tools. They'll be the ones that reshape how they work around them.Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?