Meaning Hygiene, or How Companies Turn Into Soulless Monsters

and how to resist

Every soul-crushing corporation you’ve ever worked for – the ones with meaningless meetings, projects that go nowhere, that weird hollow feeling when you ask “why are we doing this?” – they all started as startups. Passionate founders. Clear mission. Tight team that actually gave a damn.

So what the hell happens between “we’re going to change the world” and “please fill out form 27B for approval to change a button color”?

I’ve been watching this transformation up close – not just in other companies, but fighting it in my own. And I’ve realized something: there are these natural forces, like gravity, that constantly pull growing companies toward meaninglessness. Not because people are incompetent. Because humans are humans.

The Day 2 Problem (Or: Why Bezos Was Right)

Jeff Bezos has this thing he talks about in his shareholder letters – Day 1 vs Day 2:

“Day 2 is stasis. Followed by irrelevance. Followed by excruciating, painful decline. Followed by death.”

Day 1 is when you’re questioning everything, when every decision connects to mission, when “because we’ve always done it this way” isn’t even in your vocabulary. Day 2 is when operations become more important than purpose, when following the process matters more than achieving the outcome.

The dangerous part? This decline happens in slow motion. You don’t wake up one day and realize you’ve become Comcast. It happens one lazy task description at a time, one unquestioned process at a time, one “we’ve always done it this way” at a time.

I see this in companies with founders still running them (Airbnb, Nvidia, Netflix) – they manage to delay Day 2 longer. Why? Because the founder is the walking embodiment of why this exists. Even Google needed Larry and Sergey to come back for their AI Code Red, to shake everyone awake and say “the game changed, wake the fuck up.”

But here’s what keeps me up – even with a founder at the helm, there are these psychological forces working 24/7 to erode meaning. Three big ones.

Three Forces That Kill Companies Slowly

1. Cognitive Laziness Wins By Default

Here’s something I noticed: properly defining a task is genuinely hard work. It requires you to think through context, rationale, success criteria, connection to bigger goals. Your neocortex has to actually fire up and do work.

And humans are inherently lazy thinkers. Daniel Kahneman calls this WYSIATI – What You See Is All There Is. Your brain fills in missing context automatically, not realizing everyone else is filling in different context.

So what happens? Tasks start with full context: why we’re doing this, what success looks like, how it connects to our Q4 goals. Then someone’s in a hurry. They skip the “why” because “everyone knows.” Then they skip the success criteria because “it’s obvious.” Pretty soon you’re left with “make the thing” and a deadline.

The erosion is so gradual you don’t notice. Until one day you open your backlog and half of it is zombie work – nobody knows why it’s there, but nobody wants to kill it either.

This isn’t theoretical. PMI research shows companies lose about 11% of project investment to poor requirements. Standish CHAOS Report? Only 16% of projects finish on time and budget. Stripe found engineers spend 42% of their time dealing with technical debt instead of building new things.

Tasks are your atomic unit. If they’re unclear, everything built on them will be unclear.

2. Process Becomes Religion At Scale

As companies grow, someone creates a process that worked once. That process becomes The Way. Market shifts, needs change, but The Process remains because questioning it feels risky.

Following process is psychologically safe. “I followed the procedure” is an acceptable defense. Asking “should we still be doing this?” is dangerous – you might be labeled “not a team player.”

I’ve watched this happen. Good process → standard process → sacred process → process that exists long after the reason for it disappeared. And eventually you’re Comcast with Silicon Valley internet slower than what I get on a Thai island, because competitive pressure stopped mattering years ago when you achieved effective monopoly.

Netflix figured this out early – their culture document literally says “People Over Process.” Most companies drift the opposite direction until they’re drowning in forms and approvals for things nobody can remember why they needed approval.

Andy Grove said it: “Success breeds complacency. Complacency breeds failure. Only the paranoid survive.”

3. Status Quo Accumulates Silently

Quick thought experiment: look at your company’s SaaS subscriptions. How many tools are you paying for that nobody uses anymore? Tools bought for projects that ended two years ago?

This is status quo bias. In startups, there isn’t much history. But in growing companies, every past decision gains legitimacy simply by existing. Every new hire inherits “how things are done” without questioning it.

Grove had another line: “Let chaos reign, then rein in chaos.” You need both – the ability to question everything AND the ability to execute. Most companies are way too weighted toward execution without questioning.

Where This Gets Real: The Task Problem

So I did this experiment recently. Opened our task tracker, looked at random engineering tasks, asked: Could a newcomer understand how this connects to our mission?

The answer was mostly no. Not because our engineers are bad – because writing good tasks requires cognitive effort everyone shortcuts.

Look at the difference:

What most tasks look like: “Fix dashboard performance. Due: Friday”

What they should actually contain: “Why: Dashboard load time directly impacts first-day retention. When users open the app and the dashboard takes >3 seconds, they interpret this as ‘the app is broken.’ Research shows users form quality judgments in the first 50 milliseconds - after 3 seconds, they’ve decided your product feels unreliable.

The data: 23% of new users who experience >3 second load times never complete their first session. Our day-1 retention is 54% for fast loads vs 31% for slow loads - a 23 percentage point gap caused purely by performance.

This cascades: lower day-1 retention → lower day-7 retention → lower MAU. We’re losing ~200 users/month. At our LTV of $180, that’s $432K annual recurring revenue walking away because we’re technically slow. This directly blocks our Q4 goal of 15% MAU growth.

What: Get P95 load time under 2 seconds

Success:

• P95 load time <2 seconds

• Day-1 retention improves from 31% to 50%+

• No regression in data accuracy”

See the difference? The second version explains the mechanism (slow load → user thinks it’s broken → they leave), shows the metric impact (day-1 retention gap), traces the cascade to business goals, and includes the actual product metric in success criteria. The first version is just... an instruction.

Now, I know what you’re thinking: “That’s a lot more text. Who has time to write all that?”

And you’re right – it takes more effort upfront. But here’s what research on engineering motivation shows: developers perform significantly better when they understand the why behind their work, not just the what. Google’s Project Aristotle found that psychological safety and clear purpose were the top predictors of team performance. When engineers understand the business impact of their code, they make better technical decisions, catch edge cases you didn’t think of, and actually give a damn about the outcome.

The laziness problem is real – writing good context is cognitive work. But this is exactly where AI actually helps. You can draft the business context once, have it expand with the full causal chain, and suddenly that 5-minute task of writing “fix dashboard” becomes a 7-minute task that actually communicates what matters. The marginal cost is tiny compared to the cost of an engineer spending days building the wrong thing or not understanding why speed matters here.

And here’s the thing – if you can’t answer “what breaks if we don’t do this?”, you probably shouldn’t be doing it.

What We’re Actually Doing (The AI Angle)

At Welltory, my COO is rebuilding our operations around something we’re calling “meaning preservation.” The core idea: if you can’t trace a task back to mission, it’s probably waste.

We built this hierarchy:

Mission: Impact 100M people’s health measurably
    ↓
Level 1 Goals (strategic, yearly)
    ↓
Level 2 Goals (specific, quarterly)
    ↓
Projects (with 2-5 milestones each)
    ↓
Milestones (irreversible changes)
    ↓
Deliverables (things we ship)
    ↓
Tasks (actual work)

The principle: any engineer should be able to trace their task up to mission. If they can’t, something’s wrong.

But here’s what’s interesting – we’re using AI as a clarity test. Not to make decisions, but to check if we’re being clear. If AI can parse our task descriptions and understand connections, they’re probably clear. If it can’t, humans probably can’t either – they’re just better at pretending.

AI doesn’t solve the problem. It’s a mirror that shows you when you’re being vague.

Why AI Actually Helps Here

Look, I’m bullish on using AI for management. Not to replace thinking, but to scale the communication part that everyone shortcuts because they’re tired or in a hurry.

The key insight: AI doesn’t get lazy. Humans skip writing context because it’s cognitive work. AI doesn’t care. You can encode all your quality criteria into a prompt, and unlike a human reviewer, it won’t forget to check them or let things slide because it’s Friday afternoon.

Here’s the actual prompt we use to check if our tasks are meaningful or just corporate theater:

You are reviewing a task description for clarity and completeness.

Check:
1. Does it explain WHY this matters? (business context, user impact)
2. Is there a clear causal chain from this work to a product metric?
3. Can you verify completion objectively?
4. Would an engineer understand what problem they’re solving?

If any of these are missing or vague, explain what’s unclear.
Rate 1-10 for task quality.

That’s it. Paste your task, get feedback. The AI forces you to be clear enough that a machine can parse your logic. And if a machine can’t understand your reasoning, your team probably can’t either – they’re just too polite to say so.

The Real Fight: It’s Not About Perfection

Here’s what I’ve learned: you can’t stop these forces completely. Cognitive laziness, process ossification, status quo bias – they’re fundamental to how humans operate. Fighting them isn’t a project you complete. It’s ongoing resistance.

In Welltory, we have this thing we call our “mycelium network” – the web of connections between mission, goals, projects, tasks. It used to exist entirely in the heads of maybe 5-6 people (me, our CTO, a few product leads). They’d run around explaining to everyone how things connected. It was exhausting and didn’t scale.

Now we’re externalizing it. Making it visible. Not because we’re control freaks, but because the alternative is that knowledge dies with whoever leaves or burns out.

We do these strategic syncs every 1.5 months. Not yearly planning sessions where you make a pretty deck and file it away. Every 1.5 months, we ask: what still makes sense? What doesn’t? What should we kill?

People hate this at first. It feels unstable. But the alternative is worse – you wake up one day and realize half your projects are pointless but nobody knows how to stop them.

The Milestones Thing (Or: How We Track If We’re Actually Achieving Anything)

One thing we’ve gotten obsessive about: defining milestones that actually mean something.

Most companies do milestones like: “Launched beta” → “10% adoption” → “100% rollout”

These are useless. They tell you nothing about your actual strategy. They’re generic checkboxes that could apply to any project.

Better milestones tell the story of what’s changing:

❌ “Launched voice input”
✅ “25% of users who try voice input switch to using it exclusively”

❌ “Beta released to 10%”
✅ “Subjective journaling became a habit: ≥10% of weekly actives add ≥3 entries/week for 4 consecutive weeks”

See the difference? The second version tells you something about user behavior changing, about the product actually working. It’s a point of no return – something shifted in the system that you can’t (and wouldn’t want to) undo.

What This Actually Looks Like In Practice

Want to check if your milestones are meaningful or just “we launched a thing”? Here’s another prompt you can use:

You are reviewing a milestone for a product project.

Evaluate it on 5 dimensions (1–10 each):

1. Specificity — Does it describe a concrete outcome unique to this project, or a generic process any team could use?
2. Polarity — Does it have a clear opposite? (If it could describe ANY project, it’s too vague.)
3. Irreversibility — Does this milestone represent a lasting change in the system, user behavior, or business state?
4. Measurability — Can success be objectively verified (via metrics, thresholds, or observable proof)?
5. Value Link — Does it clearly connect to user or business value (e.g., retention, activation, engagement, efficiency)?

Then:
- Give a total rating (average 1–10)
- Explain what’s missing or unclear.
- Suggest how to rewrite it to make it a “point of no return” milestone.

Try it. If the AI says “this could describe any feature launch,” you haven’t defined a real milestone. You’ve just described a deployment.

The Uncomfortable Truth

Here’s what I’ve learned after years of fighting this: the transformation from startup to soulless corporation isn’t something that happens TO you. It’s something you allow through a thousand small acts of laziness.

Every time you write “fix the dashboard” instead of explaining why and what success looks like.

Every time you follow a process without asking if it still makes sense.

Every time you keep a subscription nobody uses because “we’ve always had it.”

Every time you accept “everyone knows what we’re doing” instead of writing it down.

These aren’t dramatic failures. They’re tiny acts of cognitive laziness. But they accumulate. And eventually you wake up and realize you’re Comcast.

The companies that stay alive – really alive, not just profitable-but-dead inside – are the ones that maintain what I’m calling meaning hygiene. Not as a one-time project. As a constant practice of resistance against entropy.

Because entropy is real. Cognitive laziness is real. Status quo bias is real. These forces are always working. If you’re not actively pushing back, you’re drifting toward Day 2.