Like every other ambitious owner of a growth business, we at Savvy-CFO are focused on making sense of AI's impact (and role) in our current and future operations.
Credit where credit is due: this article was prompted by Yascha Mounk’s April 25, 2026 conversation with economist Luis Garicano on The Good Fight. Garicano’s framing of “messy jobs,” and Mounk and Garicano’s discussion of “bullshit jobs,” career ladders, apprenticeship, and the future of knowledge work, helped crystallize the distinction this article builds on. The interpretation and Savvy-CFO application that follow are our own.
Artificial intelligence is changing work. And because it changes work, it changes budgets.
That much is obvious.
What is less obvious is where the real risk sits … and where the real opportunity sits. Every business needs to understand how AI affects its own work, its own people, its own training model, and its own cost structure.
That is where the distinction between clean tasks and messy jobs becomes useful.
The popular version of the AI story is usually too blunt. It asks whether AI will “take jobs.” That question is too broad to be useful. It creates fear, but not guidance. It makes young people wonder whether they should still study law, accounting, computer science, finance, marketing, design, or medicine. It makes business owners wonder whether they should hire, automate, retrain, or wait.
A better question is this:
What kind of work is AI most likely to replace, and what kind of work is likely to become more valuable?
That distinction matters.
AI is very good at work that is clean, repeatable, information-heavy, and easy to verify. It is getting better every month at drafting, summarizing, translating, coding, researching, preparing first drafts, creating PowerPoint-style material, and processing large volumes of structured information.
That does not mean every lawyer, accountant, programmer, analyst, consultant, writer, or researcher is finished.
It does mean the work inside those jobs is being unbundled.
Some tasks will be automated. Some will be amplified. Some will become cheaper. Some will become more valuable. And some career paths, especially the junior training paths that relied on routine work, will need to be rebuilt.
The key distinction is between clean tasks and messy jobs.
A clean task has several characteristics.
It is narrow. It is repeatable. It has a defined input and a defined output. It is based on accessible information. It can be checked against a known standard. It does not require much context, judgment, trust, negotiation, or lived knowledge of the organization.
Examples include:
These are not unimportant tasks. For decades, many young professionals learned their craft through exactly this kind of work.
Junior lawyers reviewed documents and drafted clauses. Junior consultants built slide decks. Junior accountants prepared working papers. Junior analysts cleaned data. Junior marketers wrote first drafts. Junior programmers wrote basic code.
That work was never only about the output. It was also part of the apprenticeship model.
The junior person produced useful work. The senior person reviewed it. Over time, the junior person learned what quality looked like, where the judgment calls were, how clients thought, how risk was assessed, and how decisions were made.
AI disrupts that bargain.
If the routine work can now be done by AI in seconds, what does the junior person offer in exchange for training?
That is one of the most important career questions of the AI era.
A messy job is not a single task. It is a bundle of responsibilities.
It involves people. It involves judgment. It involves incomplete information. It requires trust. It requires context. It requires knowing when the technically correct answer is not the right answer for the business. It requires understanding what is written down, but also what is not being said.
Messy jobs often include clean tasks. AI can help with those. But the job itself is not clean.
A good CFO role is a messy job.
Yes, parts of finance are highly automatable: bookkeeping classification, report drafting, reconciliations, variance analysis, dashboard preparation, forecasting support, research, tax summaries, and memo preparation.
But the real CFO function is not “produce the spreadsheet.”
The real work is understanding the business, identifying what matters, challenging assumptions, translating numbers into decisions, aligning leadership, managing trade-offs, anticipating risk, and helping the owner decide what to do next.
That is not a clean task. That is a messy job.
The same is true across many professional roles.
A lawyer is not only someone who drafts a clause. A strong lawyer understands the client, the negotiation, the leverage, the risk tolerance, the commercial objective, and the human dynamics around the deal.
A consultant is not only someone who builds a deck. A strong consultant helps a team see what it has been avoiding, creates alignment around hard choices, and moves people from analysis to action.
A manager is not only someone who assigns tasks. A strong manager understands people, timing, friction, capability, incentives, and trust.
AI can support all of that.
But it cannot fully replace the human who carries responsibility for the decision.
The common advice to young people used to be: build technical skills.
That advice is not wrong. It is just incomplete.
Technical skills still matter. In fact, they may matter more, because AI gives technically capable people more leverage. A person who understands finance, law, coding, marketing, operations, or medicine can use AI far more effectively than someone who simply prompts a tool and accepts the answer.
But technical skills alone are no longer enough.
The more AI can generate plausible answers, the more valuable it becomes to know which answers are good, which are dangerous, which are incomplete, and which are irrelevant to the actual situation.
That is judgment.
Judgment is not the same as opinion. Judgment is the ability to apply knowledge in context.
It asks:
AI can help answer those questions. But a person still has to know which questions to ask.
The most serious issue for young professionals is not that all work disappears.
It is that the early rungs of the career ladder may weaken.
Many professional careers have historically relied on an apprenticeship model. Young people entered at the bottom, did routine work, received feedback, absorbed context, and gradually became capable of higher-level judgment.
If AI takes over much of the routine work, firms may hire fewer juniors. Or they may hire juniors but expect them to arrive at a higher level. Or they may fail to train them properly because the traditional training work has disappeared.
That is a real risk.
It is also a warning.
Young people should not assume that a credential alone will carry them. They should seek training environments where they are exposed to real problems, real clients, real decisions, and real accountability.
They should take hard courses. They should learn how to think, write, calculate, argue, test, and defend a position. They should learn how to use AI, but not as a substitute for understanding. They should build a reputation for diligence, reliability, and judgment.
The signal of competence will become harder to fake over time precisely because AI makes polished output easy to produce.
A perfect-looking memo will not prove much.
Trust will matter more.
So will demonstrated ability.
There is another useful distinction in this debate: the idea of “bullshit jobs.”
The phrase is provocative, but it captures something many people recognize. There is work inside organizations that looks performative, unnecessary, political, or detached from real value creation.
AI will not automatically eliminate that work.
In fact, AI may increase it.
Why? Because AI is excellent at producing the appearance of work. It can generate reports, summaries, plans, dashboards, market scans, strategy documents, and meeting notes at extraordinary speed. Some of that output will be valuable. Some of it will be noise.
The production of polished nonsense is about to get much easier.
That matters for business owners.
The danger is not simply that AI replaces work. The danger is that AI floods the organization with work-like output that nobody has properly judged, challenged, or tied to a decision.
This is where leadership matters.
A business does not need more documents. It needs better decisions.
It does not need more dashboards. It needs clearer signals.
It does not need more meetings. It needs alignment, accountability, and movement.
Some meetings are waste. But some are where commitment happens. Some conversations that look inefficient are where trust is built, resistance is surfaced, and decisions become real.
That is why AI cannot be treated as an output machine alone. It has to be embedded into the operating system of the business.
For someone planning a 50-year career, the guidance is not “avoid AI.”
That would be terrible advice.
The guidance is:
Use AI aggressively, but build yourself into the kind of person AI needs beside it.
That means choosing work that is:
It also means avoiding the comfort of becoming merely a producer of clean deliverables.
A young accountant should not aspire only to prepare schedules. They should learn how the schedules connect to the business.
A young lawyer should not aspire only to draft clauses. They should learn why the clause matters, who is taking the risk, and what commercial purpose it serves.
A young marketer should not aspire only to write content. They should learn the buyer, the market, the offer, the journey, and the numbers.
A young analyst should not aspire only to build models. They should learn which assumptions drive the decision and whether the model reflects reality.
A young entrepreneur should not aspire only to automate. They should learn where automation creates value, where it creates risk, and where human trust remains the constraint.
The future does not belong to people who ignore AI.
It belongs to people who can combine AI with judgment.
The same lesson applies to businesses.
The winners will not simply be the companies that buy AI tools.
The winners will be the companies that redesign work around them.
That means identifying which tasks are clean and repeatable, then deciding whether they should be automated, accelerated, reviewed, or eliminated.
It also means identifying which jobs are messy and making those people more capable, not less.
A good AI strategy should ask:
This is not an IT project.
It is a management project.
And for many growing businesses, it is becoming a finance project as well, because every AI decision eventually turns into a resource allocation decision: where to invest, where to cut, where to hire, where to train, where to measure, and where to place accountability.
At Savvy, this is why we have built the Savvy-Lab.
Savvy-Lab is not an AI novelty project. It is not a place where we play with tools for the sake of appearing current.
It is our research and development engine.
Its purpose is to understand how business work is changing, how AI can be used responsibly, which tools and techniques best serve growth companies' needs, and how we can turn that learning into better advice, better systems, better diagnostics, and better outcomes for ambitious business owners.
We are not aiming to replace judgment with AI.
We are aiming to strengthen judgment with AI.
That distinction is critical.
AI can help us research faster, draft faster, compare options faster, detect patterns faster, and build better starting points. But the value still comes from asking the right questions, knowing the client, understanding the business model, interpreting the numbers, and helping owners make decisions in the real world.
That is why our work is rooted in ceaseless research.
The market is changing every day. AI capability is changing every day. Business systems are changing every day. Buyer behaviour is changing. Labour markets are changing. Career paths are changing.
Standing still is not a strategy.
Neither is chasing every new tool without understanding the work it is supposed to improve.
For Savvy, the answer is not to panic, and it is not to wait.
The answer is to build the capability to keep learning.
That is the practical optimism in this moment.
AI will automate clean tasks. It will expose weak thinking. It will flood businesses with polished output. It will put pressure on old training models. It will make some work cheaper and some work obsolete.
But it will also make strong people stronger.
It will make judgment more valuable. It will make trust more valuable. It will make cross-functional understanding more valuable. It will reward people and businesses that can operate in the messy middle between information, people, systems, money, and decisions.
That messy middle is where real business happens.
And that is exactly where the next generation of business leaders and their advisors should learn to work.
Next Steps
If you are trying to understand where AI belongs in your business, start with the work itself.
Which tasks are clean? Which jobs are messy? Which outputs drive decisions? Which outputs merely create activity?
Savvy-CFO helps growing businesses make those distinctions, then turn them into practical operating decisions.