Stop Selling Code. Sell the Business Outcome. (I Just Got Hired Without Applying)

Last week I started a new developer job. Here's the strange part: I never applied. There was no public posting to apply to. A recruiter found me on LinkedIn and reached out — because of the brand I'd been deliberately building since January, when I turned my portfolio site into Code350 for exactly this reason.

That's one data point in a job market that's gotten genuinely weird. Gartner projects half the companies that cut customer service staff over AI will be rehiring those roles by 2027 — Ford already rehired engineers it cut, and Commonwealth Bank reversed its AI layoffs inside a month, calling them an "error." The 2026 layoff wave is less about jobs disappearing and more about jobs being renamed while hiring pipelines keep matching on the old labels. (I wrote up the full pattern here.)

Everyone's advice for developers in this market is some version of "lean into product" or "lean into design." I think that's aiming too low. Lean into the business.

Code production is deflating. Outcome translation is appreciating.

Here's the uncomfortable arithmetic: as AI gets better at producing code, your value as a producer of code falls. Not to zero — but the floor is dropping every quarter, and more people and companies are doing their own development or pulling it in-house because the tools finally let them.

What those people and companies cannot do is the translation layer: listening to what a business is actually trying to accomplish and turning it into a working technical system. Increased productivity. Increased revenue. Fewer man-hours. Lower cost of operations. The developers who can hear one of those goals and come back with a proposal and then the deployed thing itself — those are the ones who'll be fine no matter what the titles do. Strip the buzzwords off the renamed roles in this market and that's what nearly all of them are asking for.

That's the real difference from the "learn product" advice: product-lean still positions you as a feature-maker with better taste. Business-lean positions you as an outcome-owner. Nobody relabels away the person who owns outcomes.

And there's a bigger prize hiding behind it. The most portable thing an engineer owns isn't a language or a framework — it's systems thinking. Feedback loops, bottlenecks, failure modes, state, second-order effects: that's just how we're trained to see. A business is a system. Operations are a system. A P&L is a system with particularly emotional stakeholders. The developer who can point that reasoning at an ops workflow or a cost structure — find the bottleneck, propose the fix, ship the instrumentation — can be placed almost anywhere in a company, including rooms that have never hired an engineer before. Code was only ever one application of the skill.

What my week actually looks like (and why a non-developer can't do it)

A lot of my new job is working with Claude Code. I say that without embarrassment, because the value was never the typing.

The speed comes from fluency that's invisible until it's missing: knowing your way around GitHub, hosting platforms, auth solutions — the difference between a business question spiraling into a three-week research project and it becoming an immediate answer or a seamless integration into an existing workflow. AI can write the function; it can't know that your company's SSO setup makes option B a landmine, or that the "quick" approach creates a compliance problem in month three.

One example from my own work: I built a multi-step processing queue that routes steps across different AI models based on what each step actually needs. Knowing how to integrate AI is a skill. Knowing to optimize which model handles which step turned it into a financial benefit — measurably cheaper per run. But the thing that made any of it matter was understanding the purpose of what I was building, because purpose is what shapes a pile of API calls into a product. And the product is the thing the business sees.

Developers are the best AI users alive. That forks two ways.

The "learn AI skills" advice is everywhere — OpenAI itself says the real bottleneck is skills, and the relabeling wave backs that up: the renamed roles are overwhelmingly asking for people who can take AI from demo to production. Evals, cost control, fallbacks, the unglamorous 80%.

But here's what the advice misses: developers already are the best AI users there are. We use these tools all day, against real systems, with real consequences. That advantage forks into two viable identities:

1. The skill-builders. People and companies are desperate to actually learn this stuff — not prompt-tip listicles, but real capability. Developers who can teach, build curriculum, or run enablement inside companies are selling the scarcest thing in the market: transferred competence.
2. The tool-builders. Every business is trying to operate better with AI and mostly failing at the integration layer. Developers who build the internal tools, the automations, the queues and pipelines — the things companies actually run on — are converting AI fluency directly into the outcomes from section one.

Both paths cash in the same underlying asset: you understand what these tools can actually do, and you can bolt that to a business purpose. Pick one deliberately instead of drifting.

Operate as a brand, because the search direction flipped

The part of my hiring story worth generalizing: I didn't get found by an application. I got found because there was something to find — a body of public work with my name on it that made the case before any conversation happened.

The search direction in hiring has flipped. Increasingly it's not you hunting postings; it's recruiters — and more and more, recruiters' AI tools — hunting for you. That means your public surface is your resume: the projects, the writeups, the shipped things. Titles compress badly and, as covered above, they're now actively unreliable. A visible track record of outcomes doesn't compress at all. (This is also why I built QueryQuarry, a talent marketplace where recruiters' AIs search opted-in candidates directly — full disclosure, my product — but the principle stands wherever you choose to be findable: make the machine-readable version of you exist, on your terms.)

Operating as a brand sounds exhausting if you imagine it as content-grinding. It isn't. It's making sure the work you already do leaves public evidence: the repo, the postmortem, the tool you open-sourced, the writeup of the queue you optimized. One honest artifact a month beats a hundred applications a night — I've now lived both halves of that comparison.

---

The relabeling economy is disorienting: you can lose a title that took a decade to earn and watch the same work reposted under new words a quarter later. But the confusion is symmetric — the hiring systems are as lost as the candidates — and confused markets reward whoever adapts their signal first. Stop selling code production; it's deflating. Sell the translation of business purpose into working systems, prove it in public, and let them come find you.

It works.