AI can help engineers move faster through coding, testing and exploring unfamiliar software, but it cannot decide what a business system should do or take responsibility for changes in production. That is the central argument in Tom Allen’s 22 September 2026 profile of software engineer Oleg Morgoch in The AI Journal: use AI to improve defined engineering work, while people remain accountable for the problem, the design and the result.
Who is Oleg Morgoch?
Allen’s profile describes Morgoch as a Ukrainian software engineer with nearly 20 years of experience working on legacy production systems built on Microsoft’s .NET platform. It says his work has included software for U.S. companies in real estate, oil and gas, and healthcare administration, across roughly a dozen projects. Those are biographical claims reported by the profile; they are not independently verified here.
As an Amazon Associate I earn from qualifying purchases.
The profile focuses less on a particular product or implementation than on Morgoch’s approach to engineering with AI: use the tools to assist with well-defined work, but do not outsource engineering judgment.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteWhat AI can contribute to software work
Morgoch says tools such as GitHub Copilot can assist with coding, testing, exploring code, and generating ideas for tests—particularly when an engineer gives the system a clear task. He compares AI to a driver’s navigation system: it can help find a route and suggest alternatives, but the driver still chooses the destination and remains responsible for the journey.
That distinction matters in business software. A suggested change may look plausible or even compile and pass existing tests while still breaking a rule the system has accumulated over years. The profile argues that experienced engineers must assess whether generated code fits the architecture, respects less-visible business requirements and is safe to deploy.
Allen’s article presents this as Morgoch’s perspective, not as the result of a controlled productivity study. It reports no measured project outcomes establishing how much faster his team worked or whether AI improved software quality.
Why legacy business software needs context
The profile describes older systems as a source of operational friction: accounting workflows may be disconnected, reconciliation may be manual, records may remain on paper, and invoice data may be entered by hand. These examples help explain why modernization is appealing, but the article supplies no independently sourced statistic for how common they are or how much they cost.
Recommended Free Tools
In such systems, a change is not just a coding exercise. Existing software may encode business rules that are poorly documented or understood only by the people who use it. AI can help an engineer inspect or change the code, but it cannot safely infer every operational consequence from a prompt alone. The engineer has to connect the proposed change to the real workflow and verify what happens beyond the code itself.
How to evaluate AI in an engineering process
Morgoch recommends beginning with a specific process and a measurable baseline or goal. Rather than asking how many programmers AI can replace, his proposed question is: “Which development stages can we make faster and better with the help of AI?”
- Choose one bounded process. Identify a recurring activity, such as a particular kind of code exploration, testing or application processing. Avoid treating “software development” as one undifferentiated task.
- Record the starting point. Decide what to measure before introducing the tool: for example, elapsed time, cost, output quality or the amount of human review required.
- Set a target and trial the workflow. Make clear what improvement would count and use the AI tool on the selected process, with engineers checking the work.
- Compare results against the baseline. Evaluate whether the change improved speed, cost or quality, and account for review effort and risks rather than counting generated code alone.
The profile offers illustrative goals—not achieved results—including examining a process that takes 200 hours a month, reducing an example application-processing time from 15 minutes to two minutes, or raising a hypothetical classification-accuracy target from 82 percent to 95 percent. These figures are scenarios Morgoch proposes teams might investigate; they are not reported savings, measured gains or external benchmarks.
What human review still needs to cover
AI assistance does not eliminate the need for sound software engineering. In Morgoch’s account, architectural oversight is especially important: generating code without it can contribute to disorganized systems and technical debt. A useful review therefore asks more than whether code runs.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Business behavior: Does the change preserve the rules and exceptions the organization depends on?
- Architecture: Does it fit the system’s design, or introduce avoidable complexity and maintenance work?
- Testing: Do tests cover relevant behavior and edge cases, including cases the prompt may not have mentioned?
- Production impact: Could the change affect data, connected processes or users in ways that the code-level result does not reveal?
- Accountability: Is an engineer prepared to explain, approve and support the change rather than treating the AI’s output as its own justification?
These checks follow the profile’s argument that an engineer remains responsible for deciding whether a proposed fix is correct and safe, including when it appears to work.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Protecting customer information in AI prompts
Allen reports that Morgoch’s team avoids entering real customer information into AI queries and uses test data instead. That is a specific practice attributed to his team, not a complete security policy or independent assessment. Teams adopting AI should define how they handle sensitive information and make sure their chosen workflow follows applicable organizational rules; the profile does not describe the tools’ data-handling settings or establish that any particular configuration is safe.
Adoption is also a workplace change
Morgoch observes that employees may worry about losing status, influence, control, work or job security, while managers may focus on cost. These are his observations in the profile, not findings from a broader workplace study. His proposed framing shifts the discussion from replacing programmers to identifying development stages that AI might make faster or better.
The article also quotes Morgoch’s succinct advice: “Think for yourself. Do it together with AI.” It captures the intended balance: use the tool as an assistant, while retaining human ownership of decisions and consequences.
Free tools Windows power users keep installed
One-click scans. No signup required.
What the profile establishes—and what it does not
The profile offers a practitioner’s case for applying AI to specific software tasks while keeping engineering judgment, architectural oversight and responsibility with people. It does not establish that AI has produced a particular productivity gain in Morgoch’s projects, prove that the approach works equally well in every organization, or compare coding assistants. GitHub Copilot is mentioned descriptively as one tool; the article does not test products or recommend a plan.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
