The Best AI Doesn’t Replace You. It Removes Friction.

Everywhere you look, AI is being sold as the main character.

Smarter.

Faster.

Revolutionary.

But if you sit inside real accounts and real businesses, the conversation sounds very different.

People aren’t asking:

• “Is this impressive enough?”

They’re asking:

• “Will this actually make our day easier without making us feel less human?”

That gap—between how AI gets marketed and how operators actually think—is where a lot of campaigns, launches, and funnels quietly fall apart.

This is especially true in relationship-driven businesses: clinics, agencies, expert services, vertical SaaS. Anywhere trust and judgment matter as much as efficiency.

In those environments, the best AI does one thing first:

• It removes friction so humans can do their real work.

Not the other way around.

The AI conversation we had in our call.

On paper, AI is about capability and support.

In real meetings, it’s about fear and friction.

Recently I sat in on a strategy call with a client’s product team rolling out multiple AI features for a relationship-heavy B2B product.

They weren’t just asking:

• What can this model do?

• How do we phrase the feature list?

They were asking:

• How do we talk about this so it doesn’t sound like a robot talking to our customers?

• How do we help teams do more with less without making them feel replaced?

• Which features should we ship first so people feel relief, not threat?

Underneath the roadmap, there was a very human concern:

“We want this to feel like support, not a takeover.”

That’s the real AI brief most businesses have but rarely write down.

The real job of AI: remove friction, not humans

When you look at AI through an operator’s lens, the question isn’t:

“What can this automate?”

It’s:

“Where are we burning time, attention, and judgment on work that doesn’t actually require it?”

In almost every account I touch, the high-leverage opportunities look like this:

• Context hunting

  • People spend time digging through history, documents, chat logs, tickets, records—just to get oriented before they can make a decision.

• Translation work

  • Experts produce dense, internal language. Someone then has to translate that into something a customer, client, or executive can actually understand.

• Manual glue

  • Data arrives in the wrong format. Systems don’t talk. Someone becomes the “glue” who copies, pastes, cleans up, and re-enters the same information in three different places.

None of that is where you want your highest-value people spending their best hours.

That is where AI actually belongs.

Not replacing judgment.

Not making final calls.

Not owning the relationship.

Just removing the friction that sits between “I know what’s going on” and “I can act on it.”

Three kinds of friction worth targeting first

When I’m looking at where AI could help in a business, I’m usually scanning for three categories.

1. Orientation friction

This is everything you have to do before you can even start the work.

You see it when:

• People scan long histories before every call.

• Leadership walks into a meeting and spends the first 15 minutes catching up.

• Senior staff say, “Give me five minutes to remember what’s going on with this account.”

AI is useful here when it can:

• Surface a short, accurate summary of what matters most.

• Highlight trends and outliers without hiding raw data.

• Reduce the time between “open the record” and “I see the story.”

Good implementation: humans still own the decision. They just walk in prepared instead of digging.

2. Communication friction

This is where expertise gets stuck in translation.

You see it when:

• Reports are technically accurate but unreadable.

• Follow-up emails keep getting postponed because “I don’t have the energy to write that right now.”

• Clients or customers leave a touchpoint feeling confused instead of clear.

AI is useful here when it can:

• Turn structured notes into a clear draft in plain language.

• Keep subject-matter experts in control of edits and final send.

• Help teams say what they already know—without writing from scratch every time.

Again, the point isn’t to let a model speak for you.

It’s to stop burning human energy on the first draft.

3. Integration friction

This is the boring glue work: moving information from one place to another.

You see it when:

• Teams manually re-enter data from PDFs, exports, or screenshots.

• Important details live in documents “on the side” of the system and never really become usable.

• Everyone trusts that “someone” captured the context… but nobody can find it.

AI is useful here when it can:

• Extract the right information from messy inputs.

• Structure it so it actually feeds your systems and workflows.

• Reduce the chance of “we missed that line on page four and it cost us.”

Done well, this doesn’t make the business colder. It makes it more consistent.

Why “replacement” framing backfires

Most AI marketing fails at the very first sentence.

It centers the tool instead of the relief.

Common patterns:

• “Let AI handle your [important thing].”

• “AI that talks to your customers for you.”

• “Hands-off automation for your [core relationship].”

In a spreadsheet, that sounds efficient.

In a team’s nervous system, it sounds like:

• loss of control

• loss of quality

• loss of the thing that actually makes them different

So people stall.

They ask for more information.

They keep “meaning to come back to it.”

They stay with the imperfect but familiar process they already have.

The alternative is boring—but it works:

• “Keep full control. Get a better starting point.”

• “You still approve every message. We just give you a smarter draft.”

• “Your team still decides. We remove the hunting, formatting, and busywork.”

In other words:

Don’t lead with how smart the tool is.

Lead with what it takes off their plate—and what stays human.

How to talk about AI like an operator, not a hype deck

When I’m working with founders or teams on AI positioning, I usually walk them through a simple filter:

1. What tedious job disappears?

• Be concrete. “No more 40-minute record review before every meeting.” is stronger than “smarter insights.”

2. What stays human by design?

• Spell out what the model doesn’t own: final approval, high-stakes decisions, relationship moments.

3. What gets easier for the buyer’s day, not just your ops?

• Internal efficiency matters. But people say yes when it clearly improves their experience of doing the work.

4. Where could they feel replaced—or blamed—if this goes wrong?

• Name those risks internally. Then write your copy to lower that threat level.

Most AI “strategy problems” are not model problems.

They’re messaging and decision-path problems.

You don’t need another paragraph about architecture.

You need one clear sentence about:

“Here’s what you get back: time, clarity, and fewer chances to drop the ball.”

If you’re thinking about AI in your own business

Whether you’re running campaigns, shipping features, or just trying to make your own internal systems less chaotic, the questions are the same:

• Where are your best people doing low-leverage work because the system makes them?

• Where do clients or customers feel the friction most sharply?

• Where would support be welcomed—and where would substitution feel like a threat?

Start there.

Use AI first where it quietly removes friction:

• faster orientation

• cleaner first drafts

• less manual glue

Then decide, with clear eyes, where (if anywhere) you actually want automation to take the lead.

You’ll end up with a positioning story that sounds a lot less like:

• “We added AI because everyone else is.”

And a lot more like:

• “We removed the worst 20% of your day so you can spend more time on work that actually needs you.”

If you’re testing or rolling out AI anywhere in your business and you’re not sure how to talk about it without tripping the “robot takeover” alarm, hit reply and tell me where you’re stuck.

I may turn the best questions into future pieces—anonymized, obviously—so other operators can steal the thinking.