The Friction You’re Eliminating Is the Skill You Need

TL;DR: Over-automating with AI degrades the cognitive friction that builds expertise, pattern recognition, and market intuition. When you eliminate the struggle of creating content, you lose the mental reps that develop judgment. The result: declining close rates, generic output, and hollow client interactions. The fix: use AI to augment execution, not replace thinking.

What Happens When You Over-Automate with AI

• Revenue impact: My close rate dropped from 68% to 23% in 6 weeks after full AI automation

• Lost capability: Pattern recognition and real-time objection handling disappeared because I stopped doing mental reps

• The sequence error: People try to fix trust problems (inputs) with tactic solutions (outputs)

• The recovery path: Five 45-minute client conversations rebuild market understanding faster than months of tactical optimization

• Core principle: Expertise is scar tissue from repetitions under cognitive load, not knowledge from shortcuts

Why AI Automation Fails: My Personal Experience

I was booking 8-12 sales calls per week.

Close rate: 68%.

My emails got 35-40% opens, 8-15% clicks, and 15-20 replies per send.

Then I discovered I could batch a month of emails in 2 hours using AI.

Six weeks later, my calls dropped to 2-3 per week. Close rate fell to 23%. My replies went from 15-20 to maybe 3-4.

The problem wasn’t my offer. The problem wasn’t my market. The problem was what I eliminated when I removed every moment of cognitive resistance.

Everyone’s diagnosing AI adoption as a productivity problem. Too slow, not enough ROI, poor implementation. MIT research suggests around 95% of generative AI pilots in companies are failing to deliver on their promises. Research from the Stanford Social Media Lab and BetterUp Lab suggests AI tools can create more work, not less.

The problem isn’t adoption.

The problem is what you’re eliminating when you remove every moment of cognitive resistance.

I’ve also watched this pattern with clients who automate themselves into a specific kind of exhaustion. The flatness that comes from never having to think through anything hard.

This isn’t about AI being bad. It’s about mistaking efficiency for capability, and the invisible cost of outsourcing the exact friction that builds the judgment you’re trying to scale.

Bottom line: AI automation without strategic boundaries eliminates the cognitive friction that builds expertise.

What Is Smoothout Territory?

The telltale sign someone has crossed from “using AI strategically” into what journalist Ellen Scott calls “smoothout territory”?

They stop thinking before they press the button.

They wake up. Check their calendar. See they need to write 3 emails today. And instead of spending even 30 seconds thinking about what they actually want to say, they fire up the bot. Hit generate. Copy. Paste. Send.

There’s a massive difference between “I know exactly what story I want to tell, what emotion I’m trying to evoke, and the specific action I want people to take, so let me use AI to help me express that faster” and “I need an email. AI, do the thing.”

The first one? You’re the director.

The second one? You’re a passenger.

When you’re a passenger, your copy starts getting this weird sterile quality. Everything sounds fine. Grammatically correct. Structurally sound. But it has no soul. No edge. No you.

Like eating unseasoned chicken breast for every meal. Sure, it’s protein. It’ll keep you alive. But nobody’s excited about it.

Key insight: “Smoothout” is burnout from eliminating challenge, not from overwork.

How Smoothout Differs from Burnout

Scott coined the term “smoothout” to describe a type of burnout that comes from overusing AI and erasing friction, challenge, and agency in the name of ease. This is similar to “boreout,” the mental health impact of being understimulated at work.

The trigger is not overwork or boredom, but reliance on AI tools in place of challenge.

The problem is how quickly we turn to AI at the first hint of difficulty, therefore leading to an increase in disengagement and negative stress symptoms.

When I was engaged with my writing, my emails were full of stories about my clients. Specific wins. Specific struggles. Inside jokes only my audience would get. References to current events in my niche.

Once I went full autopilot, all that disappeared.

My emails became generic. Still well-written, technically. But they could’ve been sent by anyone to any audience. There was nothing uniquely “me” about them anymore.

Your audience can feel that shift. They might not consciously know what changed. But subconsciously, they sense you’re not really there anymore. You’re just going through the motions.

That kills trust.

Critical distinction: Smoothout eliminates challenge; burnout comes from excessive challenge.

The Math on Cognitive Offloading: What You Actually Lose

Here’s what happened to my decision-making quality over 8 weeks:

Before Autopilot (Engaged Phase)

I would hop on sales calls and within the first 3 minutes, I could tell you exactly where that prospect was mentally. Not just what they wanted, but what they were afraid of. What they’d already tried. What objections were sitting in the back of their mind.

I’d been living in my audience’s head all month. Every time I wrote an email about a client struggle, I had to think about that struggle. Visualize it. Feel it. Articulate it.

By the time someone got on a call with me, I had already done 100+ mental reps thinking through every possible version of that person’s situation.

Prospect would say: “I’m struggling with client acquisition.”

I would immediately respond: “Let me guess. You’re getting leads, but they’re all price shopping. And the ones who do book calls keep asking for payment plans. Which tells me you’re probably attracting people through content that focuses on tactics instead of transformation. Am I close?”

Prospect’s jaw would drop.

That’s not magic. That’s pattern recognition built through engaged repetition.

After Autopilot (The Fall)

I would hop on those same calls and be generic.

Prospect: “I’m struggling with client acquisition.”

Me: “Yeah, that’s tough. So walk me through your current process.”

Before, I was leading the conversation. Demonstrating expertise through specificity. After, I was just asking questions. Like every other coach on the planet.

The numbers:

• Engaged phase close rate: 68%

• Autopilot phase close rate: 23%

I lost the ability to do real-time objection handling. When someone would throw out an objection, I could no longer immediately pull from my mental database of client stories. I’d resort to generic coach responses that sounded like they came from a script.

Because functionally, they did.

What degrades: Cognitive offloading eliminates pattern recognition, market intuition, and adaptive expertise because you stop doing mental reps.

What Does Friction Actually Build?

When you sit down to write an email from scratch, you have to wrestle with it.

You stare at the blank page. You write a hook. It feels weak. You delete it. You try again. Still not quite right. You try a third time. Maybe this one lands.

You move to the body. Now you’re trying to figure out how to transition from the hook to the point. It feels clunky. You rewrite it. Still clunky. You step away. Get coffee. Come back. Now you see it.

Three hours later, you have one email. And you’re exhausted.

But here’s what actually happened during those 3 hours:

You didn’t just “write an email.” You did 100+ micro-decisions. Each one was a small failure followed by a small correction. You tried something. It didn’t work. You felt the friction of it not working. You adjusted. You tried again.

Over and over.

Each time you did that, your brain was forming new neural pathways. Learning what works. What doesn’t. What resonates versus what falls flat.

You were building pattern recognition. Not from a textbook. From experience. From the scar tissue of messing up and then fixing it.

That’s how expertise gets built. Through repetitions under load. Through doing the thing, encountering resistance, and pushing through it.

The friction isn’t the obstacle to learning. The friction is the learning.

But when you let AI remove all that friction, when you just hit a button and it spits out a complete email and you don’t have to wrestle with anything, you’re not building scar tissue. You’re not doing the reps. You’re not encountering the resistance that forces your brain to adapt.

You’re just consuming.

Core mechanism: Friction creates the neural pathways and pattern recognition that constitute expertise.

The Sequence Error: Why Fixing Tactics Doesn’t Work

When someone’s in the degradation phase and their numbers start dropping, the first thing they do is look at the conversion mechanism.

“My emails aren’t converting.”

“My sales calls aren’t closing.”

“My offer isn’t compelling enough.”

So they start tweaking the outputs. Better subject lines. Stronger CTAs. More scarcity. Better closes on calls. Revised pricing.

This feels logical because in their mind, the problem is mechanical. “If I just find the right words, if I just add the right urgency, if I just structure the offer better, then people will buy again.”

They’re treating it like a math problem where you just need to find the correct formula.

But the conversion mechanism was never the problem. The conversion mechanism is downstream. It’s the output of a process.

The real problem is upstream. At the input level.

What Is the Correct Sequence?

The actual sequence should be: Market Understanding → Strategic Thinking → Tactical Execution

But what people do is skip straight to the end of the sequence. They’re trying to fix outputs without checking if their inputs are corrupted.

My script was working fine when I was connected to my market. The same exact script. What changed wasn’t the script. It was my understanding of the person on the other end.

When I said the words in my script, they were loaded with context. When I said “I understand where you’re struggling,” I actually understood. So my tonality, my pacing, my energy behind those words all communicated genuine understanding.

The prospect felt seen.

In the degradation phase, when I said those same words, they were just words. There was no database behind them. No genuine understanding. No context.

Even though the script was the same, the delivery was hollow. And prospects could feel it.

They’re trying to fix trust problems with tactic solutions. And trust is not a tactic. Trust is the byproduct of understanding.

The error: People optimize tactics (outputs) when market understanding (inputs) has degraded.

The 5 Layers of AI Use That Preserve Agency

There are five distinct layers of AI use that preserve agency. Knowing which layer to operate on for each part of your process is the whole game.

Layer 1: Ideation (What to Say)

Bad version: You ask ChatGPT for 10 email ideas. It spits out generic options like “The #1 mistake preventing your success” or “Why most people fail at [thing].” You pick one. Write the email. Send it. It’s fine. But it could be for anyone in any market.

Better version: You spend 30 minutes thinking first. “What’s a recent client conversation that surprised me? What objection keeps coming up that I have a unique take on? What belief does my market have that’s actually wrong?”

You write those down. Then you go to AI and say: “I had a client who believed X, but I showed them Y, and here’s what happened. Help me turn this into 10 different email angles that explore different aspects of why this belief is limiting.”

Now the AI is working with your insight. Not replacing it.

When to use AI:

• Automate: Never. Ideation should always have human input.

• Augment: Always. Use AI to expand on your ideas, not generate them from scratch.

• Abstain: When you don’t actually have an insight. Don’t let AI fake one for you.

Layer 2: Research & Context Gathering

Bad version: You ask AI what the main pain points are for people trying to lose weight. It gives you generic responses. Lack of motivation. No time. Don’t know where to start. Technically accurate. Completely useless. Because 10,000 other marketers are getting the same answers.

Better version: You feed AI actual data from your market. Sales call transcripts. Email replies. Survey responses. Reddit threads from your niche. YouTube comments on competitor videos.

Then you say: “Analyze these transcripts and pull out the exact language people use when describing their struggle with X. Give me the specific phrases, metaphors, and emotional descriptors.”

Now you’re not getting generic pain points. You’re getting the actual words your market uses. Which you can drop into your copy and create instant resonance.

When to use AI:

• Automate: The processing of large amounts of data. Let AI find patterns.

• Augment: The interpretation. You decide what patterns matter.

• Abstain: When you don’t have real data. Don’t let AI make things up.

Layer 3: Structure & Frameworks

Bad version: You give AI your idea and it writes a complete email. You read it once. “Good enough.” Copy. Paste. Send. You just let AI make every structural decision. Where to put the hook. How to transition. When to introduce the CTA. What tone to use. You’re a passenger.

Better version: You engineer the structure yourself. “I want to open with a provocative statement that challenges conventional wisdom. Then I want to tell a 3-sentence story about a client who believed that conventional wisdom. Then I want to reveal the one shift that changed everything for them. Then transition to why most people can’t make that shift without help. Then CTA to book a call.”

You give that to AI as the blueprint. It fills in the blanks. But you made the strategic decisions. You decided the emotional journey. You controlled the experience. AI just handled the sentence-level execution.

When to use AI:

• Automate: The sentence construction. The actual writing.

• Augment: Let AI give you multiple structural options. You choose.

• Abstain: Never let AI decide your structure without your input. Structure is strategy.

Layer 4: Voice & Style

Bad version: You tell AI to write in a casual, conversational tone. Every AI user says this. So you get the same homogenized “casual” voice. Starts sentences with “Look.” Uses phrases like “Here’s the thing” and “At the end of the day.” It sounds like an AI trying to sound human. Because it is.

Better version: You train AI on your actual voice. You feed it 10-15 emails you wrote. Or transcripts of you talking. You identify your specific patterns. “I use short sentences for emphasis. I swear strategically, not randomly. I make jokes at my own expense. I use pop culture references. I break the 4th wall and call out what I’m doing.”

Then you say: “Write this email in my voice. Use my sentence rhythm. My humor style. My specific verbal tics.”

Now it’s not “casual.” It’s you. And there’s a difference.

When to use AI:

• Automate: Never fully. Always edit for voice.

• Augment: Use AI to write drafts. You punch up the personality.

• Abstain: First drafts. Write the first version yourself to establish voice. Then let AI help with variations.

Layer 5: Editing & Refinement

Bad version: You generate an email. You read it once. “Looks good.” Send. No editing. No iteration. No improvement. You’re trusting AI’s first output is optimal. (Spoiler: It’s not.)

Better version: You generate an email. Then you run it through a checklist. “Does this open with a hook that creates curiosity? Is there a specific story or example? Do I show instead of tell? Is the CTA benefit-driven? Would I personally want to read this?”

You identify 2-3 weak spots. Then you use AI strategically. “Rewrite this opening to be more provocative.” “Give me 5 different ways to illustrate this point with a metaphor.” “Punch up this CTA to be more compelling.”

You’re using AI as a sparring partner. Not a ghostwriter.

When to use AI:

• Automate: Running copy through objective quality checks.

• Augment: Generating alternatives for weak sections. You choose the best.

• Abstain: Final approval. You have to be the final filter. Always.

Framework summary: Use AI to augment execution at each layer, but never replace human strategic thinking.

How to Recover: The 5-Conversation Method

When I realized I was in smoothout territory, my instinct was to do everything at once. “I need to rewrite all my emails! I need to research my market! I need to update my offer!”

That was panic mode. And panic mode is how you stay stuck.

Here’s the actual first move: Have 5 conversations with recent clients.

Not prospects. Not people who didn’t buy. Clients. People who actually paid you money in the last 90 days.

The Specific Question to Ask

“What was going on in your life or business right before you decided to buy from me?”

Not “why did you buy.” Not “what do you think of my program.” What was happening before.

That question takes them back to the moment when they were still in the problem. Before your solution. Before the transformation. When they were still struggling.

When they describe that moment, they’re going to use current language. The words they actually think in. Not the sanitized, post-purchase rationalization.

What You’ll Hear

You’re going to hear the specific frustrations. The exact fears. The internal dialogue. The failed attempts. The moment they knew something had to change.

While they’re talking, you’re not just listening. You’re absorbing. You’re rebuilding your mental database in real time.

You’ll have 3-5 “oh no” moments where you realize: “Wait, they’re not worried about what I thought they were worried about.” “Oh, they’re using completely different language now.” “This objection I spent 6 months addressing? Nobody cares about it anymore.” “This is what’s actually keeping them up at night.”

Just like that, your understanding starts coming back online. The database starts repopulating.

Why This Works

You’re not doing this to gather “content.” You’re not recording it to feed into AI. (Though you can do that later.) You’re doing this to reconnect. To remember what it feels like to actually care about understanding someone. To rebuild the muscle of listening. Not listening to respond. Listening to absorb.

This works because it’s the smallest viable action that has the highest impact on rebuilding your mental model.

You don’t need to talk to 50 people. You don’t need to read 1000 survey responses. You don’t need to do a whole market research project.

Just 5 conversations. 45 minutes each. That’s 4 hours total.

In one afternoon, you can start reversing months of atrophy.

What Happens After

After you do this, everything else gets easier. When you sit down to write an email, you’re not staring at a blank screen. You’ve got 5 real stories in your head. 5 real struggles. 5 real transformations.

When you hop on a sales call, you’re not guessing what to say. You know what resonates. Because you just heard it. In their own words.

When you’re trying to come up with a hook, you don’t need AI to make something up. You just remember what Sarah said last Tuesday about how she felt like she was “spinning her wheels in mud.”

That’s your hook. “Ever feel like you’re spinning your wheels in mud?”

Instant connection. Because it’s real.

The recovery protocol: Five 45-minute client conversations rebuild market understanding in 4 hours.

What Changes When You Fix the Input Problem

I called up a recent client. Had the conversation. When I hung up, my whole energy was different.

“I’ve been talking about the wrong problems for 3 months. Everything I thought they cared about, they don’t. The thing they’re actually struggling with? I never even mentioned it.”

That one conversation gave me more clarity than 3 months of “optimizing” my tactics.

That night, I wrote an email. Manually. No AI. Just me and what I learned from that conversation.

It was the best email I’d written in 6 months. Because it was real again. It had soul. It had me in it.

My audience felt the shift. I got 40+ replies to that email. People saying: “Yes. This is exactly what I’m dealing with.” “Where have you been? I’ve missed emails like this.”

After one conversation. One email. The reconnection had begun.

The Real Cost of Automation

Let’s say you “saved” 15 hours per month by going full autopilot. Over 3 months, that’s 45 hours saved.

But if your revenue dropped from $60K per month to $15K per month, that’s a $135K revenue hit.

You “saved” 45 hours and it cost you $135K. That’s $3,000 per hour you “saved.”

And now you have to spend 3-6 months rebuilding, which is probably another 100+ hours of focused work, to get back to where you already were.

In total, you’re 145 hours in the hole and $135K down. All because you wanted to batch emails in 2 hours instead of 8.

The hidden cost: Saving 45 hours through full automation can cost $135K in lost revenue plus 100+ hours to rebuild.

Why Expertise Is Scar Tissue, Not Knowledge

Most people think skill-building works like this: Learn the information, apply the information, you’re skilled.

So they consume courses. Read books. Watch videos. “Now I know how to write copy. Now I know how to close sales. Now I know how to run a business.”

But then they go to do the thing and it doesn’t work. Or it works but not the way they expected. Or it works once but they can’t replicate it.

They’re confused. “But I learned the framework! I followed the steps! I know the theory!”

You have knowledge. But you don’t have expertise.

Because expertise isn’t built through learning. It’s built through friction.

Expertise isn’t knowledge. It’s scar tissue.

How Intuition Actually Forms

When you’ve done the reps, when you’ve built the scar tissue, when you’ve wrestled with 100 different emails and encountered 100 different problems and solved those problems through iteration, you develop intuition.

The ability to sense what’s needed without having to consciously think through every step.

You can look at a piece of copy and just know: “That hook’s going to fall flat.” “That transition’s too abrupt.” “That CTA needs more benefit.”

You don’t have to analyze why. You just feel it. Because your brain has done the reps. It’s pattern-matched against thousands of micro-experiences.

But when you skip the friction, when you let AI do all the wrestling, you never build that intuition.

So when you encounter a new situation, something the AI wasn’t trained on, something that requires adaptation, you’re stuck. Because you don’t have the foundational capacity to figure it out.

You only know how to press the button. And if the button doesn’t give you the right answer, you have nothing.

Where Growth Actually Happens

Growth only happens at the edge of capability. Not beyond it. (That’s just chaos.) Not below it. (That’s just comfort.) Right at the edge. Where you’re capable of doing the thing but it requires effort. Where there’s friction.

That’s the zone where your brain goes: “This is hard. I need to adapt.”

And adaptation is growth.

But if you remove all the friction, you’re never at the edge anymore. You’re just coasting. In the comfort zone. And no growth happens there.

You’re trading short-term efficiency for long-term dependency. You’re getting faster results at the cost of building actual ability.

And you won’t even realize you’re doing it until the moment you need that ability and it’s not there.

Core principle: Expertise forms through repetitions under cognitive load, not through consuming information.

How to Use Friction Strategically

Use AI to remove the tedium, not the tension. Because tedium is waste. But tension is growth.

You don’t need to manually format your emails. That’s just grunt work. Let AI handle that. You don’t need to rewrite the same sentence 10 times to fix a typo. That’s not building skill. That’s just tedious.

But the thinking? The wrestling? The part where you don’t know what to say and you have to dig for it? Keep that. That’s where the growth happens. That’s where expertise gets built.

What to Automate vs. What to Preserve

Automate these (tedium):

• Formatting and layout

• Fixing typos and grammar

• Data processing and pattern identification

• Generating multiple options for comparison

Preserve these (tension):

• Strategic thinking about what to say

• Understanding your market’s current state

• Deciding structural and emotional flow

• Final judgment on quality and authenticity

In a world where AI can do most things, the only sustainable edge is human capability. Not just knowing what works, but having the adaptive capacity to figure out what works in situations you’ve never seen before.

And that capacity isn’t built through shortcuts. It’s built through struggle. Through friction. Through scar tissue.

The hard way is the only way. Everything else is just borrowed capability. And borrowed capability disappears the moment the lender leaves.

Strategic use: Eliminate tedious tasks with AI, but preserve cognitive tension that builds adaptive capability.

Frequently Asked Questions

How do I know if I’ve crossed into smoothout territory?

The telltale sign is that you stop thinking before pressing the generate button. If you’re using AI without first spending time clarifying what you want to say, what emotion to evoke, or what action to drive, you’re likely in smoothout territory. Other symptoms include declining engagement metrics, generic output that lacks your voice, and inability to handle unexpected situations on sales calls.

Can I reverse the damage from over-automation?

Yes. The fastest recovery method is having 5 conversations with recent clients (people who paid you in the last 90 days). Ask them: “What was going on in your life or business right before you decided to buy from me?” These 45-minute conversations rebuild your mental database in about 4 hours total. Then write one email manually based on what you learned.

What’s the difference between smoothout and burnout?

Burnout comes from excessive challenge and overwork. Smoothout comes from eliminating challenge through over-automation. Smoothout is the mental health impact of being understimulated because AI has removed the cognitive friction that creates engagement and skill development. Both lead to exhaustion, but from opposite causes.

Which parts of my workflow should I actually automate with AI?

Automate tedious tasks that don’t build capability: formatting, fixing typos, processing large data sets, and generating multiple options for comparison. Preserve cognitive tension that builds expertise: strategic thinking about messaging, understanding market context, deciding structure and emotional flow, and making final quality judgments.

How long does it take for over-automation to damage my results?

Based on my experience, noticeable degradation begins around weeks 5-8. By week 8, you’ll see declining reply rates and conversion metrics. Your pattern recognition and market intuition start degrading because you’ve stopped doing mental reps. The longer you stay on full autopilot, the more expensive the recovery becomes.

Why does my AI-generated copy feel generic even though it’s well-written?

Because AI generates content based on patterns, not authentic understanding. When you’re disconnected from your market, AI fills the gap with generic responses that could apply to anyone. Your copy lacks specificity, current market language, inside references, and your unique voice. The audience subconsciously senses the lack of genuine understanding, which erodes trust.

What’s the actual ROI calculation on AI automation?

In my case, I “saved” 45 hours over 3 months by automating email creation. But my revenue dropped from $60K/month to $15K/month—a $135K loss. That’s $3,000 per hour “saved.” Recovery took 3-6 months and 100+ hours of focused work. The real cost isn’t time saved; it’s capability lost and revenue declined.

How do I use AI without losing my expertise?

Use AI to augment execution, not replace thinking. Follow the 5-layer framework: (1) Always generate ideas yourself, use AI to expand them; (2) Feed AI your real market data, don’t let it invent insights; (3) Engineer the structure yourself, let AI handle sentence-level writing; (4) Train AI on your voice, then edit heavily; (5) Use AI as a sparring partner for refinement, but you make final decisions.

Key Takeaways

• Cognitive friction builds expertise: The struggle of creating content generates pattern recognition, market intuition, and adaptive capacity. Eliminating this friction through over-automation degrades these capabilities.

• Close rates dropped 45 percentage points in 6 weeks: My conversion rate fell from 68% to 23% after full AI automation because I lost the ability to demonstrate specific understanding and handle objections in real time.

• The sequence error is fixing outputs when inputs are corrupted: When results decline, people optimize tactics (better copy, stronger CTAs) instead of rebuilding market understanding. Trust is built upstream through genuine understanding, not downstream through better words.

• Five client conversations rebuild market intuition in 4 hours: Ask recent clients “What was going on right before you decided to buy?” to rebuild your mental database faster than months of tactical optimization.

• Use the 5-layer AI framework to preserve agency: Never automate ideation. Feed AI real data for research. Engineer structure yourself. Train AI on your voice but edit heavily. Use AI for refinement but make final decisions.

• Expertise is scar tissue from repetitions under cognitive load: Knowledge comes from consuming information. Expertise comes from wrestling with problems through friction. You can’t shortcut the neural pathway formation that creates intuition.

• Automate tedium, preserve tension: Let AI handle formatting and data processing. Keep the strategic thinking, market understanding, and creative wrestling that builds capability.

– Razvan

P.S. If you’re realizing you’ve been on autopilot and your results have dropped, don’t try to fix everything at once. Text 5 recent clients today. Say: “Hey, I’m doing some research to better serve my audience. Would you be open to a quick 30-minute call where I ask you about your experience before working together?” Schedule those calls this week. Have the conversations. Listen. Take notes. Then write one email based on what you learned. Manually. No AI. Just you and the keyboard. Send that. Watch what happens.