What "Good at Work" Means Now

For ten years, every project I ran followed the same cycle. Plan, design review, dev review, QA, ship. The topics changed — different products, different teams, different problems — but the cycle never did. I thought I was free. The freedom was inside the cycle. The cycle itself was fixed.

That's what working inside a workflow feels like. You can move within the structure, but you can't change it. When the structure itself is the routine, working better just means moving slightly faster within the same five steps.

The cycle wasn't there because the work needed five steps. It was there because we needed to see where things broke. Was the bottleneck in design or in QA? Splitting work into steps made failure visible. And since redoing a cycle was expensive — weeks, sometimes months of rework — the elaborate workflow wasn't overhead. It was insurance.

AI changed the math. Throw a problem at a model and you have a draft in minutes. Don't like it? Tell it what's wrong and run it again. The cycle that used to need months of upfront planning now takes an afternoon. If it's wrong, you start over for free.

When that happens, the elaborate workflow stops being insurance. It's just overhead.

So what replaces it? Programmers already had a name for the answer.

Programmers split their tools into two kinds: libraries and frameworks. The difference is who's in charge. With a library, you are. You design the flow, write the code, and reach for the library when you need a specific function.

A framework works the opposite way. It already has the flow. You fill in the parts where judgment is needed — route handlers, form validation, business logic. Spring, React, Django are all the same idea. The framework runs the show, and you supply the parts only you can write.

Programmers call this Inversion of Control. The same flip is happening at work.

The workflow era was the library model. You designed the whole flow. You called specialists at each step. You made the judgment calls between stages. Planner calls designer, designer hands off to developer, developer passes to QA. You were the engine.

In the framework era, you set up the structure — purpose, audience, constraints, what counts as good — and AI runs the cycles inside it. Your job moves from running the cycle to designing it. From doing each step to deciding what gets done.

Take a marketer. In the workflow era, the job was: research the trend, write five drafts, review with the team, revise, publish, run an A/B test. Each step was a human step.

In the framework era, the job is to define the frame. Who's this message for — first-time signups, or returning users we've lost? What words can we never say — competitor names, medical claims, hype superlatives? What counts as a win — opens, clicks, or conversions above 5%? What triggers a legal review — health claims, financial promises, mentions of minors? Once the frame is set, AI runs the cycle — drafts fifty versions, checks each one against the frame. The human only steps in when something hits a red line.

Investors face the same flip. The workflow era was: read the deck, run the numbers, take the meeting, write a memo, debate it at partners' meeting, decide. Every deal walked the same path.

In the framework era, the investor sets the frame. What's the thesis — where do we think value compounds in the next decade? What's a deal-breaker — burn rate, founder background, market size? What's the hurdle for a yes? AI screens hundreds of decks against it, surfaces the ten that pass, drafts the first-round memos. The investor steps in for the judgment calls AI can't make — reading the founder, weighing what's not in the deck, deciding when the frame itself needs to change.

The reason this matters isn't elegance. It's reusability. Workflows had to be rebuilt for every project — different domain, different team, different tools. A good framework keeps its shape across projects. Only the contents change. A workflow is a recipe; a framework is a principle of cooking. The recipe gets you one dish. The principle works on any dish.

One good framework replaces dozens of workflows. Which is why "good at work" is starting to mean something different. It used to mean running cycles efficiently. Now it means designing them well.

Programmers didn't invent the framework idea. Architects had it first — the frame goes up before the walls, the walls before the rooms. Each layer sets the rules for what comes next. Software just borrowed the pattern and renamed it. Most useful frames already exist. They're just sitting in another field, or another era, waiting for someone to carry them over.

Workflows didn't disappear. They just moved. Running them isn't the work anymore. Designing the framework is.