I guess I’m an engineer now

For decades, access to production was guarded by a craft boundary. You had to live inside the editor before you were allowed near the live system. That boundary is thinning. Today, you can describe desired behavior in natural language while the platform brings the pieces you used to wire by hand: sign-in, infrastructure, and the connections to other systems. Assistants no longer just autocomplete a line; they map work, reshape code, save milestones, and keep going for long stretches. The bottleneck moves to diagnosis and constraint-setting. That is where domain experts already live.

If you need a label for the new workflow, “vibe coding” is the popular one. The name is imperfect, but the direction is right: you state intent, test what comes back, and iterate in dialogue rather than by stitching files. You still need judgment. You still need to know what good looks like. The difference is that the heavy lift has moved from keystrokes to framing, critique, and iteration.

The road from intent to deploy is vanishing. The skill is steering the conversation that makes the product real.

There is another current running through the stack that matters even more for organisations: application building is melting into the data plane. Look at how the Hera Space Companion sits on Azure AI Foundry with Data Agents and Microsoft Fabric. You converse with a live mission and the plumbing that once demanded a dozen services is now part of the same surface where you ask questions and push updates. The lines between “the app,” “the platform,” and “the data” are blurring in ways that favor people who understand the business, the data model, and the user’s moment.

This is where launches like Lovable Cloud fit. Not as hype, but as a sign that standing up a database, adding accounts, and wiring integrations have moved into the same conversational flow as the product description. The point is not one release. The point is that the distance between intent and deployed software keeps shrinking.

Score the release, not the origin. Users care about outcomes, not authorship.

Culture will determine whether the shift pays off. Harvard Business Review documents a competence penalty: reviewers rate identical work as lower quality if they believe an assistant helped. It feels rational in the moment, especially in code review. It is not. It slows adoption where it would yield the most, and it punishes exactly the behavior that makes assisted teams effective. Judge outcomes, not provenance. Normalize assistance where it raises quality or reduces lead time. Blind the obvious choke points for bias. Publish the deltas that matter and move on.

Who shows up in this new reality? Recent OpenAI data suggests the early gender gap in ChatGPT usage has narrowed substantially; by mid-2025, names classified as typically feminine make up just over half of weekly users. That aligns with what many leaders notice in practice: a style of working that treats the model as a partner, tests ideas in dialogue, and moves quickly from rough structure to running behavior. Pair that with the earlier evidence base on emotional intelligence at work and you get a simple conclusion: the skills we once called “soft” are now the hard edge that makes these systems productive. Train for them. Reward them. Make them ordinary.

Soft skills are the hard advantage now. Teach them, reward them, make them routine.

None of this erases engineering. It relocates it. The craft does not vanish; it becomes the set of capabilities a team draws on when the work demands it. Shipping software is becoming less about who is allowed to type and more about who can keep a high-quality conversation tethered to reality. Many of them will sit closer to the business than to the terminal. Many of them, if usage data is any guide, will be women. That is not a slogan. It is an operating assumption for how to hire, teach, and lead in the next few years.

If this clarified the terrain, use it. Quote it. Share it. Drop it into your next deck. And if you do, feel free to point people here. That is how useful thinking travels.

Curated Source Material

This piece comes from the field. If you want the research layer, here is a curated set of sources that frame the patterns discussed.

• Introducing Lovable Cloud and AI, Lovable, Sep 29, 2025.
• What is vibe coding?, IBM; Google Cloud explainers, 2025.
• Research: The Hidden Penalty of Using AI at Work, Harvard Business Review, 2025.
• How people are using ChatGPT, OpenAI, 2025.
• Hera Space Companion, Microsoft Documentation, 2025.
• Emotional intelligence – the essential skillset for the age of AI, Capgemini Research Institute, 2019.