In the late 1960s, elite Navy pilots began losing dogfights.
The deep, instrument-level understanding of exactly where they were, what their aircraft was doing, and what was coming next had been automated. And when moments of crisis arrived, they didn’t have the situational awareness to respond. Put a plane on autopilot long enough, and the pilot stops actually flying.
The same dynamic is playing out across enterprise software. AI is generating code faster than developers can understand it, and leaders are celebrating the velocity without asking who’s actually flying the plane.
A developer who has only ever “vibe coded” has perception at best. They can “see” the outputs but can’t fix any internal failures caused by the very AI systems they’re relying on. The easiest thing to do is to say the answer looks good enough. Cut and paste it in and hope it works out. According to Model Evaluation & Threat Research’s randomized control trials, experienced developers working with AI tools actually took 19% longer to complete tasks than those working without them, despite predicting beforehand that AI would make them 24% faster.
The fundamentals of good software delivery have never been more important — and never more neglected.
When instruments go dark
The Navy’s answer to training dogfighters for success was the Top Gun school — not just to teach pilots to fight, but to teach them how to fly again. That meant returning to the fundamentals by mastering the technical and combat skills that can best prepare them for moments of crisis with clear thinking. This very discipline makes split-second decisions possible when everything is on the line.
Consider this scenario. A retail company’s engineering team used AI to refactor a promotions engine ahead of the holiday season. The code passed every test. Reviews were clean. It shipped on a Tuesday with zero flags.
But what if nobody caught that AI had subtly changed the order of operations in a discount calculation? It’s a logical shift that wouldn’t have broken any individual test case but could compound incorrectly when multiple promotions applied to the same cart. This would be enough to cost the company millions by the time a finance analyst notices the margin erosion during a quarterly close.
The vibe coding wave is already breaking. One analysis found that roughly 10,000 startups tried to build production apps with AI assistants; more than 8,000 now need rebuilds.
It’s on us to turn this moment of reckoning into an opportunity.
Training developers for real-world applications
So what can we as leaders do about this?
The foundation of everything we do comes down to trust — specifically, teaching people to trust and own their work. My high-level vision is to help people achieve a 50x improvement in their overall processes by leveraging AI tools and still being the expert at large.
For one of Copado’s internal programs, we gave nine employees the ability to vibe code AI-powered tools to tackle any major business problem they identified. Most gravitated toward the same theme: they were constantly fielding repeat questions and wanted to stop answering the same thing twice.
But while the instinct was right, the execution wasn’t ready. They hadn’t thought through who would maintain these tools, how they would be governed, or whether they actually mapped to business objectives.
Just because you can hand someone the controls doesn’t mean they know how to fly.
We then conducted a training session on how to plan an app effectively — with a long-term view of the full software development life cycle — before anyone wrote a line of code. The app ideas got sharper, and the products got real.
The group went from pursuing 10 app ideas to a focused set of seven, with two participants stepping back after realizing they didn’t yet have a problem worth solving. Five are now being implemented across the business: Legal built a policy bot to answer HR’s questions on company policy; the doc writing team built a tool for automatically generating technical documentation; the support team built a case analysis app; the sales team built a call-coaching app that helps sales development reps improve performance by analyzing live calls; and the customer success team built an app that listens to calls and notes, then automatically summarizes everything known about a new client at the point of implementation.
To this day, we also reserve “Failure Fridays,” a monthly space for employees to practice debugging programs without AI assistance. It keeps foundational skills sharp and ensures that when something breaks in production, the team knows how to actually fix it.
Five pillars for AI applications
Across a community of 120,000 Copado developers, I now recommend they enforce these five pillars when deploying AI in their projects:
- Build in checkpoints to evaluate agent output against defined standards before anything moves forward.
- Continuous and automated testing should function as a permanent trust layer embedded directly into the development cycle.
- Apply human judgment at critical decision points while automation handles the routine verification work in between.
- A single review at the end of a process is a point of failure. Continuous validation is necessary to catch issues the moment they arise rather than after they’ve compounded.
- Maintain audit trails and performance metrics that capture every agent action. Accountability means tracking what AI does, not just what developers deliver.
I believe that success demands the technical knowledge and discipline to build these systems from the ground up. These guardrails ensure that AI works with you, not against you. The bottom line: organizations that approach AI with accountability and knowledge in mind achieve 9x to 10x productivity while maintaining trust.
At Copado, fostering a culture where developers are genuinely motivated to embrace AI is equally important to us. To support that, we created a certification and incentive program that rewards new hires with $1,000 bonuses upon completion — an investment that has delivered a 76% ROI compared to traditional onboarding methods. The impact has been undeniable: we had 30 developers fully onboarded in just 30 days, condensing what typically takes three to six months into a fraction of the time.
The fundamentals will endure
Speed without situational awareness isn’t efficient. It’s a deferred crisis.
The fundamentals of planning, building, testing, and releasing aren’t bureaucratic overhead — they’re the instruments on the dashboard, telling you where you are, what your system is doing, and what’s coming next. Lose them, and you’re not just flying blind. You’re unprepared for the dogfight.
When the moment of reckoning arrives — the production failure, the security breach, the audit, the outage — you find out very quickly whether a human’s full understanding was there or not.
The machine won’t be in the hot seat. You will.
—
New Tech Forum provides a venue for technology leaders—including vendors and other outside contributors—to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to doug_dineley@foundryco.com.
