Don't Scale Slop

The most dangerous thing you can do with AI is automate a broken process. You don't get a better process. You get broken at 10x speed.

This is the mistake most business owners make when they get excited about AI: they skip straight to "how do I automate this?" without first asking "is this process actually good?"

Growth Multiplies Everything (Including the Bad Stuff)

There's a pattern that repeats at every scale. You hit a revenue milestone and it feels like winning. At the same time, your best people are leaving, your clients are frustrated, and the internal systems are held together with duct tape.

This is not a coincidence. Growth does not fix broken systems. It multiplies them. If your onboarding process loses 30% of new clients to confusion, scaling that process with AI means you lose 30% of clients faster, at higher volume, with less human judgment catching the errors.

The same applies to:

• A sales process that closes deals but creates mismatched expectations
• A content workflow that produces volume but not quality
• A customer support system that responds quickly but doesn't actually resolve issues
• An internal communication pattern where decisions bottleneck at the founder

AI makes all of these faster. It does not make any of them better.

The BuzzFeed Warning

BuzzFeed has $185 million in annual revenue. Their market cap? $26 million. That's a 0.14x revenue multiple. For context, YouTube (now the largest media company on earth at $62B+ revenue) trades at 8-9x.

What does 0.14x mean? Investors are saying: this business is going extinct.

BuzzFeed built an empire on algorithmically optimized content at industrial scale. Listicles, quizzes, shareable formats. They were extraordinarily good at the game they were playing. The problem is the game ended. AI can now produce infinite slop faster and cheaper than any human content farm. When anyone can generate a thousand BuzzFeed articles in an afternoon, the entire business model of "capture attention, sell ads" breaks down. The supply of content goes to infinity. The value per unit goes to zero.

BuzzFeed scaled execution. They did not scale substance. They never graduated from producing content to building systems that could adapt when the ground shifted. They never invested in the human elements that AI cannot replicate: taste, trust, conviction, the kind of work that makes people stop scrolling and sit with it.

This is the cautionary tale for every business owner thinking about AI. If you use AI to produce 10x the content, all optimized for today's algorithm, all dependent on today's distribution channel, you are building a faster BuzzFeed. And the market has already told you what that's worth: 0.14x.

The companies that will win are the ones building at a higher level. Systems that can adapt. Talent pipelines that create real value. Brands that audiences trust independent of any single platform. Substance that AI cannot manufacture.

Where You Sit in the Value Hierarchy

BuzzFeed was what we'd call a Player organization: elite at execution, but never graduated to coaching (designing systems that could adapt). There's a broader framework for understanding the levels of value in the AI economy, from Spectator to Player to Coach to Game Creator to Game Engine Creator. The minimum viable position is Coach. If you're not at least designing the systems your business runs on, you're vulnerable to the same compression that killed BuzzFeed.

The full framework is in The Five Levels of Value in the AI Age. For business owners, the takeaway is simple: AI is compressing execution toward zero. The cost of taste, trust, domain expertise, and leadership didn't collapse. Those are the assets that appreciate. Everything in this article is about how to build on them instead of scaling slop.

The Founder Bottleneck

Before you can scale anything, you need to ask: am I the bottleneck?

Every business is capped at the founder's personal capacity. This shows up in predictable ways:

• People wait on you to make decisions before they can move forward
• You do 10-20% of every team member's job because "it's faster if I just do it"
• You are the junk drawer of the business: a little bit of everything, nothing clearly owned

The pattern of growth reflects this:

• $0 to $1M: You learned how to sell
• $1M to $3M: You learned how to delegate (a little)
• $3M to $10M: You learned how to hire people who can actually do things
• $10M+: You need to learn how to lead. This is where most founders stall.

The traits that made you successful early (moving fast, having all the answers, controlling everything) become liabilities at scale. Moving fast becomes chaos. Having all the answers prevents your team from thinking. Controlling everything makes you the bottleneck.

The endgame is clear: the founder should be at zero percent of the bottleneck for daily operations. Zero. The founder's job at scale is purely strategic: new business lines, key hires, partnerships, vision. If you're still in the weeds of daily execution, you haven't built the systems. You've just built a job that requires you to show up every day.

The question is not "what do I need to change about the business?" It is "who do I need to become?"

Three Levels of Business Infrastructure

That question has a concrete answer. It starts with what your business actually runs on.

Level 1: Documentation. You write things down. Processes exist in Google Docs. Checklists live in Notion. It sits there until someone opens it and follows the steps. This is where most businesses stop. It feels organized. It is not a system.

Level 2: Triggered Workflows. You trigger a process and it runs. When a new client signs, onboarding kicks off automatically. When a content piece is approved, distribution fires. The human initiates, the system executes. This is a meaningful upgrade, but it still depends on someone remembering to pull the trigger.

Level 3: Autonomous Operations. The system acts on schedule or in response to conditions, whether you remember or not. Your morning briefing generates itself before you wake up. Your content pipeline identifies trending stories, assigns them to creators, and handles pre-production without anyone opening an app. Quality checks run on their own. Drift detection catches problems before they become crises.

Most businesses never reach level three. They have great setups, great templates, great checklists. But nothing happens unless someone opens the app and starts clicking. The system has zero agency of its own.

AI makes level three achievable for businesses that previously needed armies of middle managers. But only if the processes at levels one and two were actually good. If you automate broken documentation into autonomous operations, you get autonomous slop: a system that produces bad output at scale, on schedule, with nobody watching it happen. That's not just worse than manual slop. It's the most expensive kind of slop there is, because it compounds while you sleep.

When you do reach level three, the human role changes completely. You're no longer operating the machine. You're fine-tuning it. Watching the outputs, adjusting the inputs, improving the quality of the autonomous system over time. The business becomes an autonomous value creation machine, and the founder's job is to make that machine better, not to run it. This is the transition from Player to Coach in practice: you stop doing the work and start improving the system that does the work.

What Battle-Tested Looks Like

Before you automate a workflow, it should pass these checks:

1. Can you draw it? If you can't draw the workflow as a linear sequence of steps (see Workflow Decomposition), you don't understand it well enough to automate it. Automating something you can't draw means encoding confusion.

2. Has a human done it successfully, repeatedly? The workflow should have been executed by a person (or team) enough times that you know what "good" looks like. You need a baseline. Without a baseline, you have no way to measure whether the AI version is better or worse.

3. Are the inputs and outputs clearly defined? Every step should have a specific input (what triggers it) and a specific output (what it produces). If a step is "review and improve," that's not defined enough. What are you reviewing? What does "improved" look like in observable terms?

4. Is the decision logic documented? When a human makes a judgment call in the workflow, what criteria are they using? If the answer is "they just know," you need to extract that knowledge before you automate. Otherwise the AI will guess, and it will guess differently every time.

5. Do you have feedback loops? When the workflow produces a bad outcome, how do you know? How quickly? If there's no feedback mechanism, you'll scale errors silently.

The Accountability Readiness Test

Your team's readiness to work with AI mirrors their general accountability level:

Level	Description	AI Readiness
1	Can't start without being told what to do	Not ready. AI will create more chaos, not less.
2	Can do tasks but can't make decisions without input	Barely ready. AI can handle rote tasks only.
3	Can do tasks and get feedback before finishing	Getting there. AI can draft, human reviews.
4	Can do tasks, make decisions, get feedback after	Ready. AI handles execution, human handles exceptions.
5	Can do tasks, make decisions, no loop-in needed	Fully ready. AI becomes an autonomous agent within clear boundaries.

If your team is at Level 2 and you deploy Level 4 AI automation, you'll have AI making decisions that nobody on the team knows how to evaluate. That's not automation. That's abdication.

The path: train your people up the accountability dial first. Then bring in AI at the level your team can actually supervise.

The Right Sequence

1. Know where you are. Which level of value are you operating at? Which level of infrastructure does your business run on? Be honest.
2. Fix the process first. Map it, test it, measure it. Make it work with humans. You cannot skip this step.
3. Document the decision logic. Extract the judgment calls into observable criteria.
4. Train your team to the right accountability level. They need to be able to evaluate AI output.
5. Then automate. Start with the lowest-risk, highest-volume steps. Expand from there.
6. Use the time AI frees up to climb. Move from Player to Coach. From Coach to Game Creator. From daily operations to pure strategy. Every hour AI saves you is an hour you can invest in higher-level work.

This sequence is slower than "just plug in AI and see what happens." It's also the only sequence that doesn't end in scaled slop.

The Bottom Line

AI is compressing execution toward zero. The value is moving upward: from Player to Coach to Game Creator. Taste, trust, domain expertise, leadership. These are appreciating assets in a world where the technical elements get cheaper every quarter.

Systems are not for when your company gets big. They are prerequisites for getting big. If you skip them, you're not scaling a business. You're scaling a mess.

AI is the most powerful scaling tool ever created. That's exactly why you need to be honest about what you're feeding into it. Fix the process. Build the infrastructure. Climb the levels. That's how you scale something worth scaling.

See also: The Five Levels of Value | Quick Check | Situation Map | Workflow Decomposition | Beyond Automation