[{"data":1,"prerenderedAt":18},["ShallowReactive",2],{"post-why-dirty-work-deserves-serious-consideration":3},{"slug":4,"title":5,"excerpt":6,"content":7,"tags":8,"readingTime":15,"publishedAt":16,"published":17},"why-dirty-work-deserves-serious-consideration","Why dirty work deserves serious consideration","There is a category of work that looks like work, fills the calendar, and produces no creative output. It comes back every day, in slightly different forms, and never quite goes away. It is called dirty work, and it is the silent killer of every team that is supposed to be doing creative work. The system — or the agent — is the only thing that can move it out of the way.","\n## Dirty work, named\n\nThere is a category of work that looks like work. It fills the calendar. It produces visible artifacts — a spreadsheet, a status update, a CSV import, a UTM-tagged link. It is the work that comes back every day, in slightly different forms, and never quite goes away.\n\nI call it dirty work. Not because it is unimportant. Most of it is, in fact, structurally important — the operations of any team depend on it. It is dirty because it consumes the hours of the people who are supposed to be doing the work that the team actually needs: the creative work.\n\nThe dirty work, by definition, is the work that could be done by a system. It is repetitive. It is well-defined. It has clear inputs and outputs. It does not require taste, judgment, or a sense of the customer. A competent agent, given the right context, can do it. A backend service, given the right schema, can do it faster.\n\nThe creative work, by definition, is the work that cannot be done by a system. It requires a point of view. It requires a judgment call. It requires the kind of taste that only forms when a person has spent years inside a problem.\n\nThe two are not equal. The creative work is what the team is for. The dirty work is what the systems are for.\n\n## How the trap closes\n\nHere is how the trap closes.\n\nYou hire a creative person — a writer, a designer, a marketer, a strategist. They are good at creative work. You give them a brief. They produce something good. The team is happy.\n\nThen a small task appears. \"Can you add UTM parameters to this link?\" It takes five minutes. They do it. Then another: \"Can you import this CSV into the system?\" Ten minutes. Then another: \"Can you check the analytics for last week's campaign?\" Twenty minutes. Then another: \"Can you set up the waitlist for the new program?\" An hour. Then another: \"Can you re-export this asset at three aspect ratios?\" Thirty minutes per asset. Then: \"Can you sync the contacts between the two systems?\" An hour. Then: \"Can you write the shipping tracking template?\" A morning.\n\nNone of these tasks, individually, is unreasonable. All of them are real work that needs to be done. The person who does them is competent at them, often quicker than the next person. So the work goes to them. And the next task goes to them. And the next.\n\nAfter six months, the person is spending sixty percent of their week on dirty work. The creative output — the writing, the design, the strategy — is what gets squeezed into the cracks. The person is no longer being paid to do what they are good at. They are being paid to be a competent generalist, which is the lowest-value use of a creative specialist.\n\nThe trap closes quietly. The team did not decide to make the creative person a generalist. It happened, one task at a time, until the role was defined by the dirty work and the creative work was the side project.\n\n## What it costs\n\nThe cost of the dirty work trap is not visible in any single quarter. It shows up over years, in three places.\n\n**Creativity dies first.** The person who used to ship one new concept per week now ships one per quarter, because the time is gone. The work that does get shipped is more conservative — the person does not have the cognitive bandwidth to take creative risks when they are tired from the dirty work. The team's output gets safer. The team's output also gets less interesting. Customers notice, slowly.\n\n**Retention suffers second.** The person is hired to be creative. After two years of mostly doing dirty work, they take a job somewhere else where they get to be creative. The team loses the institutional knowledge. The team also loses the next hire, who hears from the alumni that \"the work is mostly operations.\"\n\n**Growth stalls third.** The team scales linearly with headcount, because the dirty work scales with output. A team of three creative people produces three times the output of one. A team of one creative person plus three dirty-work contractors produces one times the output of the previous team, plus the cost of the contractors. The team never compounds.\n\n## What dirty work actually looks like\n\nI keep using the phrase \"dirty work\" as if everyone knows what it means. Here is a concrete list, drawn from the work I have actually seen and done. Each item is real. None of it requires taste. All of it can be done by a system, an agent, or a backend service — and the moment the system is in place, the person should be doing something else.\n\n- **UTM links.** Every campaign needs UTM-tagged links. There is a utility for this. It takes seconds. It should not be a person's job.\n- **Waitlist setup.** Every new program needs a waitlist. There is a form. There is an email sequence. There is a CRM integration. None of it requires a creative decision. All of it requires a system.\n- **Shipping tracking.** Every order needs a tracking template, a status update, a customer email. This is operations. It is well-defined. It is a backend service.\n- **Data analysis.** The dashboard exists. The data exists. The query is the same query, run every week. The analyst should be reading the result and making decisions, not running the query.\n- **Data sync.** Two systems, two schemas, one source of truth. This is a data engineering problem. It has a solution. It is not a person's job.\n- **Analytics.** The events fire. The pipeline runs. The dashboard renders. The marketer should be looking at the dashboard and making decisions, not maintaining the pipeline.\n- **Marketing operations.** The content goes out. The distribution happens. The engagement is tracked. The marketer should be making the content and the strategy, not doing the distribution by hand.\n\nEach of these is, on its own, a real task that someone competent can do. None of them, on its own, justifies a person's week. Together, they consume every week.\n\n## The principle\n\nHere is the principle. It is the inverse of the observation, stated positively.\n\n> Dirty work belongs to the system. Creative work belongs to the person.\n\nA team is a system of people whose job is to do the work that a system cannot do. A backend service is a system whose job is to do the work that a person should not be doing. The two should not be confused. The two should not overlap.\n\nWhen a person is doing dirty work, one of two things is wrong. Either the system has not been built yet, or the system has been built but the person does not trust it to do the job. In the first case, the answer is to build the system. In the second case, the answer is to fix the system until the person does trust it.\n\nThere is no third case. There is no \"we just need one more person to handle the operations.\" That is how the trap closes.\n\nIn the AI agent era, the \"system\" that does the dirty work is not always a backend service. Sometimes it is an agent. A review agent that triages requirements. A coding agent that writes the diff. A form that forces the user to think about the edge case before they file it. The principle is the same: the dirty work goes to whatever can do it without losing the creative judgment, and the creative work stays with the person.\n\n## What I am not saying\n\nThe principle is not a recipe for over-automation. Some work feels like dirty work but is actually creative work in disguise. The line is not \"automate everything that is repetitive.\" The line is \"automate everything that does not require taste.\"\n\n**Judgment under uncertainty is not dirty work.** \"Should we run this campaign?\" is a question that requires taste, knowledge of the customer, and a willingness to be wrong. It is not dirty work. It is the work. The person should be doing it.\n\n**Strategy is not dirty work.** \"What should we be doing in the next quarter?\" is creative work. So is \"what should this page say?\" So is \"what should the customer feel when they open this email?\" The dirty work is the production of the strategy, not the strategy itself.\n\n**Sales is not dirty work.** The first call with a customer is creative work. The follow-up is creative work. The negotiation is creative work. The CRM updates, the pipeline reports, the quota tracking — that is dirty work. The dirty work should be in the backend. The creative work should be in the room with the customer.\n\n**First drafts are not dirty work.** A junior copywriter's first draft is creative work, even if it is rough. An agent can clean it up, but the agent cannot produce the point of view that the first draft is trying to articulate. The principle is to put the system around the creative work, not in front of it.\n\nThe line is not \"do less work.\" The line is \"do the work that requires taste, and let the system do the work that requires repetition.\"\n\n## Where this connects\n\nThis is the same principle I keep returning to, in every system I have built. Each system is a different surface, applied to the same insight.\n\n**The GUILD Digital Operations Ecosystem** is this principle applied to the social media operation. The SMMS, the LINKS wall, the AURORA module, the Field Expedition flow, the tracking layer — every one of those modules exists so that the editors, the strategists, the overseas operations leads can spend their week on the work that requires a human and not on the work that requires a system. Planners stopped chasing statuses in chat. Editors stopped retyping the same caption in five places. Designers stopped re-exporting the same asset at three aspect ratios by hand. The systems did it, and they did it the same way every time.\n\n**The GUILD Bio Link Surface** is this principle applied to the link in the bio. The featured banner is config-driven. The post grid is a SQL join that decides which posts are public-visible. The UTM tagging is a utility. Every piece of the page that used to be a person's job is now a system's job. The marketing team can launch a campaign in five minutes, because the system did the dirty work for them.\n\n**The GUILD Certificate Intelligence Ecosystem** is this principle applied to the moment of verification. The QR code, the certificate page, the cross-brand conversion path, the analytics layer — every piece of the page that used to be a manual process is now a system. The person who used to reformat PDFs is now working on the next exhibit. The person who used to send confirmation emails is now working on the next gem of content.\n\n**The Agent Pipeline** is this principle applied to the requirement-to-PR loop. The structured form, the review agent, the coding agent, the architect — every step that used to be a developer's job is now an agent's job. And more importantly, every step that used to be the bottleneck for a sales lead filing a real requirement is now an agent's job. The sales lead, the operations lead, the developer — all of them get to spend their week on the work that requires a human, and not on the work that requires a system.\n\nEvery one of these systems is the same insight, applied to a different surface. The insight is: dirty work is the enemy of creative work, and the system — or the agent — is the only thing that can move it out of the way.\n\nThe hardest part is not building the system. The hardest part is recognizing that the dirty work is being done by the wrong person. And then, instead of hiring another person, building the system that lets the existing person do the work they were hired to do.\n    ",[9,10,11,12,13,14],"Process","AI Agents","Team Design","Org Design","System Thinking","GUILD",8,"2026-08-06",true,1785988758943]