Recently, I saw a CTO vibe-code a compelling web application on their weekend, secure the enthusiastic support of the C-suite the following week, and then think that production readiness will follow, with one developer, before the end of the month.
A subsequent estimate of two to four months was met with surprise. The gap is not art; it is the reality of the best technological delivery: quality and security stability, visibility, compatibility, data management, performance and operational readiness.
The article continues below
UK and Ireland’s Engineering Leader, Slalom.
The “we don’t need engineers” soundbite makes the belief that AI tools can replace the technology team as a whole, rather than produce powerful results.
We hear some boards honestly assert that in six months they won’t need an engineering team at all, as some companies ditch engineers to rededicate resources to AI-focused tasks and AI products.
But this undermines the human part of the role and will ultimately diminish organizations once they realize that programs that coded the vibe have already failed or will soon go away.
This article delves into why production-ready software still needs the support of engineering teams, despite the rise of the coding vibe – in fact the accelerated development of AI is a deep technical endeavor in and of itself.
The role that AI plays in acceleration
There is no doubt that AI is reshaping the way we build software. We’ve been using it across the spectrum (from rapid testing to multi-team, multi-year projects) and it’s embedded across the SDLC (Software Development Life Cycle). Not just in a single developer’s IDE (Integrated Development Environment) but in creating a feed and backend, UX and architecture, code and tests, deployment, and performance.
The results are strong: faster cycles, leaner teams, stronger quality signals, and better documentation. This change is happening at scale, and businesses are experiencing measurable benefits in performance, reliability and time-to-value. Therefore, we see reduced delivery times and the need for a smaller team of application developers.
‘Vibe coding’, defined as rapidly assembling prototypes and LLMs and no code/low-code tools, prepares a persuasive demo is an accepted practice across the industry, as well as in people’s free time, today. That this can be done so easily fuels the narrative that engineering teams are selective.
AI-supported engineering can compress timelines and reduce team size, but it doesn’t eliminate the need for experts, or the calendar time needed to meet production-ready requirements. In fact, it is a deeply technical work of its own.
Unpacking production readiness
Manufacturing readiness means different things to every organization, but in general a manufacturing readiness checklist should consider the following:
– End-to-end quality
– Security, privacy and compatibility
– Reliability, resilience and disaster recovery
– Awareness
– Performance & scalability
– Accessibility
– Maintainability
In regulated industries, the above in addition to auditing, management, and final tracking are table stakes and every change must be proven.
These are not optional features or finishing touches – they are the finished product. If these requirements are not defined, tested, and automated in source code, pipelines and runbooks, then the business has a prototype, not a system.
Benefits of AI and efficiency
Within those businesses at the forefront of adoption, we’ve seen clear evidence that AI is improving the end-to-end SDLC. There are quick fixes for takeovers and backlogs, sharp architecture options, rapid UX testing, amazing code and test production, and live, valuable documentation.
The result is shorter lead times and higher throughput. With this, it is true that smaller groups can send more with clear signals as there is more focus on quality. Cadence is also changing: instead of coming together in two-week sprints, teams are moving to continuous flow-based delivery with feature flags, canary releases and deep visibility.
With this change, quality engineering has become a first-class specialty because we need specialized experts who can use risk-based testing, security-based testing, automated security, performance and durability testing, and create Evals to validate our information and model results.
While ambitious CTOs and entrepreneurs see vibe writing as a way to quickly cut corners and perfect teams, in reality, AI is raising the bar for engineering success, not lowering it. The basics are more important than ever, like the core software engineering principles of DRY (Don’t Repeat Themselves) and SOLID, clean architectures with clear communication, and automated build, test, and deployment pipelines.
Compatible with live streaming
Traditional barriers to going live are disappearing. However, as developers accelerate, the entire business and the surrounding energy cannot keep up. Classic UAT (User Acceptance Testing) and SIT (System Integration Testing) cycles are slow and operations teams in most cases are not set up to support intraday change, and it is not a small effort to get there.
So, what can be done when you hear vibe-coding soundbites from leaders or customers? To balance speed and security, there’s a three-way playbook to help teams navigate building with AI:
1. Test (days) to confirm feasibility
2. Testing (weeks) to solidify architecture and establish initial SLOs (Service Level Objectives), release pipelines, security scans, and visibility
3. Production (multiple sprints) to satisfy business standards
Teams can define production-ready expectations and automate the collection of evidence so that release gates are driven by data. For example, using DORA metrics (lead time, deployment frequency, change failure rate, and Average Recovery Time) will help manage flow and reliability as it builds.
It’s important to reset all business expectations and remember that a prototype is a signal, not a schedule. Celebrate the momentum AI tools provide but avoid doing production days without a demo.
Then make sure there is a place to finance the equipment that makes the speed safer. Businesses need to continue to train or hire quality engineering, field engineering,
DevSecOps and SRE (Site Reliability Engineers) skills. Development teams in AI assisted engineering, risk-based UAT, and flow-based delivery, before delivering security, compliance, legal, support, and finance and providing the support needed for regular, small versions.
Speed ​​up with AI, but keep the quality high
AI is a real accelerator, potentially compressing build time significantly, but production readiness is still achieved through quality, safety, performance, and manageability. The way forward is not to slow down, it is to reorganize on clear roads with automatic proof at every gate – speed and safety go together.
Businesses should remember to invest where it matters, in field engineering, SRE, DevSecOps and quality engineering discipline. Taking this approach will help you leverage loyalty at speed so that the power of the demo becomes brand loyalty. Build fast, finish responsibly.
Check out our list of the best DevOps tools.
