Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Info Inlet’s claim that AI revealed “one real skill” is a personal provocation, not proof that AI has made developers’ implementation skills obsolete. The essay’s useful distinction is between production—turning a specification into working software—and judgment—deciding whether that software is correct, durable and safe for the people relying on it.
What prompted the realization
In an essay published on DEV Community on September 30, 2026, Info Inlet recounts assembling an invoice tracker with AI assistance in about forty minutes. The speed, the author says, raised an unsettling question: after a decade spent learning to build software, how much of that work was distinctive expertise rather than implementation effort that a tool could now supply?
As an Amazon Associate I earn from qualifying purchases.
That forty-minute account is the author’s own estimate, not a controlled comparison or an independent measure of what AI can build. The essay offers no external study or statistic showing that developers’ production skills have become obsolete. Its headline is best read as a reflection on professional identity, not a measured conclusion about the software workforce.
Production and judgment are different kinds of work
Production turns a specification into software
Info Inlet uses “production” for the visible craft of writing code: working with languages, frameworks and APIs to implement a requested feature. AI can make this part feel less scarce when it generates code quickly, especially if a developer has traditionally measured contribution by lines written, tickets completed or features shipped.
#1 Best Overall
Judgment asks whether the result deserves to run
“Judgment,” in the essay’s framing, is the work of looking past clean code or passing checks to ask whether the behavior is actually right in context. Does the system keep its promises when a request fails halfway through? Can a retry create a duplicate or lose an update? Will a customer be left with a success message but no usable result?
The distinction is an interpretive lens, not a hard boundary: production involves decisions, and judgment often shapes implementation. The author’s point is that both have long been bundled into a developer’s job, so automating more of the visible implementation can feel like losing part of one’s professional identity.
Why the failed write mattered
To illustrate the difference, Info Inlet recounts a past write path that acknowledged a client before a row had been saved. A retry then coincided with a failed save, and, in the author’s account, a paying customer lost access without useful logs. The incident is an anecdote from the essay, not an independently verified account.
Recommended Free Tools
The technical issue is the gap between what the response appeared to promise and what had become durable. A client that receives an acknowledgment may reasonably act as though the change succeeded; if storage has not completed, the user-facing success and the system’s actual state can diverge. Retries can add another complication: depending on how an operation behaves, repeating it may be harmless, may duplicate an effect, or may leave the caller unsure what happened.
Google Cloud’s guidance for its C++ client libraries says an operation is generally idempotent when repeating successful calls leaves the same system state as one successful call, and that only idempotent operations are safe to retry in general. It also treats retry eligibility, transient errors, duration and backoff as policy decisions—not automatic properties of every operation. Google Cloud’s retry-policy documentation provides that general context; it does not verify the anecdote or prescribe one universal order for acknowledging and saving application data.
How to review AI-generated implementation
The essay’s practical challenge is to pause before accepting plausible output and ask how it could fail in the real application. That review is more useful when tied to the promises a feature makes than to whether the generated code merely looks tidy.
Rank #4
- Define success: What does the response tell the user, and what must already be true for that promise to be accurate?
- Trace state changes: If a process stops between steps, what has been persisted, what has not, and what will the next request observe?
- Examine retries: Is repeating the operation safe? If not, how does the system avoid unintended duplicate effects or recover from an uncertain outcome?
- Make failures observable: What information will help someone diagnose a failed or partial operation without relying on the customer to report it?
- Keep a human accountable: Who decides that the behavior is acceptable for the users, data and consequences involved?
Tests can preserve a known failure scenario—for example, ensuring a retry does not repeat an effect—but they cannot automatically identify every relevant failure mode. Finding those scenarios still requires design and review informed by the system’s behavior and the consequences of getting it wrong.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWhat the essay recommends developers change
Show decisions, not just output
Info Inlet recommends describing consequential judgment calls in a CV: the risks considered, trade-offs made and reasons a particular behavior was safe enough to ship. A list of languages and completed features can show what someone has built; explaining how they handled a risky state change can show how they think.
Best Value
Stop using volume as the only measure of value
The author also argues against measuring personal worth solely by lines of code, tickets or features. Those counts can describe activity without capturing whether the work prevented a customer-facing failure or made a system dependable. That is a career perspective, not a universal rule for evaluating engineers.
Separate generation from sign-off
Info Inlet describes an agent design with an author that produces, a skeptic that challenges the output, and a human who makes the final call. The author puts the principle plainly: “So I don’t let the thing that writes the code be the thing that signs off on it.” The design is the author’s philosophy, not evidence that a particular workflow guarantees reliable software.
What “one real skill” does—and does not—mean
The essay closes with a question: “of everything you know, which single skill would still be yours if AI could do all the rest tomorrow?” It is a prompt for reflection, not a forecast that AI will perform every other part of development. Likewise, Info Inlet’s line—“Ten years didn’t build me ten skills. It built one. The other nineteen I was renting from a supplier who just took them back.”—expresses the author’s feeling about changing tools and identity; it is not a finding about every developer.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesA grounded takeaway is narrower: if a tool makes some implementation work faster, developers may need to make their less visible contributions easier to recognize. In this essay, that contribution is judgment—the capacity to question what software promises, how it behaves under failure, and whether its consequences are acceptable. Whether that is the single skill that remains valuable is a personal question, not something this one account settles.
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.
