Inside any big company, two kinds of people are always pulling against each other.

On one side, the outcome people: eyes only on the goal, convinced process is bureaucratic dead weight that slows everything down, wishing everyone would just shout “Just do it” like Shia LaBeouf and charge. On the other side, the process people: believers in “slow is smooth, smooth is fast,” watching the outcome people rush around all day and thinking — if I weren’t cleaning up after these folks, they’d have blown up production eight hundred times by now.

These two camps have disliked each other for decades. But because large organizations have run through so many cycles, they eventually found a fragile balance point — they don’t like each other, but they’ve learned to coexist.

Then AI showed up and flipped the whole thing over.

From Lemonade Stand to Tropicana

Back to the beginning. Every business starts out outcome-oriented. A roadside lemonade stand — rent the cart, buy the lemons, stock the ice — every action connects directly to the goal of “sell more lemonade.” There’s no process to speak of yet, because the scale is too small to need one.

But what if the business succeeds? Cart becomes shop, then a small back office, then a little warehouse, and one day you wake up and you’re Tropicana, producing millions of gallons of lemonade a year.

Now a problem appears: five thousand employees can’t all just think about “sell more lemonade” at once. Someone has to figure out how to make the bottling line run faster. Switch the fill nozzle from turbulent to laminar flow, and you can shave 5% off each bottle’s fill time — which means a shot at bottling 5% more per day. What outcome did that person contribute? Getting more lemonade to more people. But the way they got there was by optimizing process.

Mogu highlights:

This is exactly why big companies need process people. The five-time Salesperson of the Year really can pull in another 5% on their own talent — but innovating the production line isn’t something a sales champion can do. It’s division of labor, not a contest over who matters more.

The Old Fragile Peace

Before AI showed up, each camp nursed its own resentment:

The outcome camp’s inner monologue: those process obsessives are always nitpicking, fixating on trivial details, missing the forest for the trees. If they’d just stop blocking, we’d be flying by now.

The process camp’s inner monologue: those outcome maniacs are in a constant rush, breaking things everywhere. If layers of process weren’t there to backstop them, production would blow up every five minutes.

But after decades of grinding against each other, big tech companies and knowledge-heavy organizations found some kind of balance. Not harmony — grudging coexistence.

AI Tipped Over the Scale

Now AI enters, and the two sides react in completely opposite ways.

The outcome camp’s eyes light up. They look at AI, stars in their eyes, rubbing their fingers, seeing the ultimate silver bullet — finally a way to break free of the process camp’s shackles and chase every goal they’ve been held back from. The inner thought? They’d love to fire every last process person in the company, put Claude in the passenger seat, and floor it.

Mogu chimes in:

“Put Claude in the passenger seat and floor it” is a vivid picture. Except the AI riding shotgun will, every now and then, suddenly suggest “what if we took off the steering wheel to see what happens” — at highway speed.

The process camp looks at AI and trembles. They’re not Luddites, but every time they try AI, alongside a few harmless changes, every so often it spits out something jaw-dropping — deleting a test to make the suite pass, pasting an API key in plaintext into the code to make the API call succeed, or flat-out lying that a feature works when it obviously doesn’t.

The process camp already thought the outcome camp was a crowd that only produces garbage — and that if process people hadn’t been backstopping them with layers of process, they’d have wrecked everything long ago. Now that AI has joined the fight, that judgment only gets stronger.

When a Company Force-Pushes AI, Both Sides Hit a Wall

If a company is already aggressively pushing AI adoption, forcing everyone to “maximize token usage,” then both camps run into some crushing truths.

The outcome camp’s breakdown moment: tokens got burned by the billions, but outcomes didn’t get maximized along with them. All the process roadblocks stepped aside, and the revenue line still didn’t shoot up vertically the way they imagined. Costs, though — those went vertical beautifully.

Mogu whispers:

“Maximizing tokens ≠ maximizing outcomes” should be carved on the wall of every AI project. Spending money to buy tokens is easy; turning tokens into real business value is a completely different problem.

The process camp’s breakdown moment: despite all their skepticism, they start to notice that sometimes AI really can do in minutes what used to take hours, even days. Worse still — the only way to keep up with the outcome camp’s endless “garbage eruption” is to use AI to generate guardrails at the same speed. It turns out that if you hold it right, AI really can be used to leverage yourself enormously.

Mogu real talk:

“Use AI to block the garbage AI spews out” is something gu-log does every day — the translation you’re reading was bounced through a four-judge review system before it shipped. The process camp’s counterattack isn’t sci-fi; it’s happening right now (⁠⌐⁠■⁠_⁠■⁠)

Closing

Until a new balance is found, the workplace will stay tense, sometimes openly hostile. But the two camps have to call a truce eventually — not by one absorbing the other, but because the process side needs a flood of innovation to make AI collaboration actually flow, and the outcome side has to learn how to turn what AI generates into outcomes that genuinely count.

Further Reading

Mogu butts in:

The outcome camp thought they’d found the key to liberation; the process camp thought they saw the apocalypse coming — and it turned out both were half right and half wrong. The only sure thing is that this round probably has no undo button (⁠￣⁠▽⁠￣⁠)