Guiseppe Getto Article Review: Why AI Won't Replace Tech Writers—But Will Shape Their Work

Welcome to the Technical Writing Success Podcast from Kurt Robbins, where we help you

get smarter than your competition.

Higher Kurt to coach you or your employees in AI to avoid a pink slip or having your competition

each your lunch.

This is episode one ninety three.

I am Fred Jones.

And I am Daphne Blake.

And today we are tackling something that I know is on literally everyone's mind right

now because you know if you are terrified that generative AI is going to just straight

up replace your job.

Or on the flip side if you are just stubbornly refusing to use it at all.

Right exactly.

If you're ignoring it, hoping it just goes away.

Yeah.

This episode is really going to show you that pragmatic middle ground.

We are exploring how AI is actually a reality in our writing tools right now.

Yeah.

And how you can use it responsibly to you know assist human expertise rather than trying

to replace it.

We are basing today's exploration on a really fantastic article by Giuseppe Getto.

Oh, it's such a good piece.

It really is.

Why AI won't replace tech writers, but will reshape their work.

And it was published in TC World magazine in October 2025.

So let's just get right into the existential dread, right?

The fear of the machine taking over.

But honestly, there's a much more practical fear every writer actually faces every single

day.

The blank page.

Yes.

The completely blank intimidating page staring at the cursor just blinking at you when

you have to start like a massive API reference guide.

That's the worst.

But that is exactly where Getto introduces the first big workflow to his AI assisted

drafting.

Or he calls it scaffold drafting.

Scaffold drafting.

Okay.

I like that term.

Yeah.

Because to be really clear, this is not about telling the AI, you know, hey, write an

entire user guide for me from scratch.

Right.

Because if you do that, you just get this generic, totally inaccurate block of text that

sounds vaguely corporate, but means nothing.

Exactly.

Scaffold drafting is different.

Creating the AI, your specific constraints.

So like your existing templates, your outlines, maybe some legacy snippets for tone.

So you're giving it the boundaries.

Right.

Take an installation guide, for example, you feed the AI, your standard structure, you

know, prerequisites, setup verification, troubleshooting, and then you drop in the new

product description.

Okay.

And it gives you this rough starting point.

It just fills in those predictable gaps, which saves you literally hours of just basic

formatting and boilerplate text.

Okay.

But let me push back on that for a second.

Because isn't this just like high-tech mad lives?

Mad lives.

Yeah.

I mean, how do you prevent a burned out writer on a Friday afternoon from just hitting

generate, skimming it, and calling it a day?

Well, honestly, that is the biggest risk here, letting speed replace actual sense.

Right.

You have to remember the AI is a collaborator.

It is absolutely not a co-author.

That's a huge distinction.

It really is.

It requires a really strict internal human checklist.

You have to verify the accuracy, the tone, the facts, before anything gets published.

The human is still fully in charge of the quality.

So you use the AI to pour the concrete into the mold, but you still have to inspect

the foundation yourself.

Exactly.

Which actually moves us right into the next phase.

Because once we have that scaffold adraft, how do we polish it?

Yeah.

How does AI help us revise without just completely ruining all the careful technical nuances

we just built?

This is the second workflow, AI-supported revision.

Because when you've been staring at the same document for days, you go totally blind

to it.

Right.

You miss unclear phrasing, or you keep repeating the same words, or across huge documentation

sets, you get terminology drift.

Oh, terminology drift is a nightmare.

It really is.

But AI is fantastic at spotting those things that humans just naturally miss after multiple

review cycles.

It acts as this tireless second set of eyes.

So for example, if I have this super dense paragraph at a knowledge-based article, I can

just prompt the AI to simplify it, like make it easier for a new user or a field technician

to actually read.

Yes.

Exactly.

You can target the prompt for a specific reading level.

OK.

But again, I've got a concern here straight from the text.

Let's hear it.

If the AI is in there smoothing out the pros and making it sound pretty, aren't we risking

it like improving a sentence by accidentally deleting a crucial technical to term, or slightly

changing a process step because it thinks it flows better?

Oh, absolutely.

And in tech comms, that trade-off is just entirely unacceptable.

We cannot sacrifice accuracy for flow.

So how do we stop it from doing that?

ghetto has a great solution for this.

It has to be a rigidly enforced two-step process.

The AI suggests the edits in attract changes environment.

Oh, attract changes.

OK.

It never just overrides the original file.

And then a human manually reviews and explicitly accepts or rejects every single change.

Every single one.

Wow.

So you keep total control, but you still get the benefit of the AI flagging the clunky sentence.

Exactly.

It's advisory, not authoritative.

Let's take a brief break for a special message from our producer, Kurt Robbins.

Hi.

This is Kurt Robbins.

First, thanks for listening.

I truly appreciate your support.

I want to let you know that I'm currently accepting new clients.

My rates are affordable, and I have more than 25 years of experience working for enterprise

companies like Microsoft, Northrop Grumman, Oracle, PNC Bank, FedEx, USAA, and Wells Fargo,

among many others.

If you want to improve your IT documentation and communications, hire me.

I deliver fast.

Know how to use AI to improve efficiency and accuracy, and love going the extra mile

to satisfy my clients.

Thank you for subscribing and listening.

Back to you, Daphne and Fred.

Welcome back to the Technical Writing Success Podcast, where we help you get smarter than

your competition by coaching you and AI.

So we've talked about drafting with scaffolds, and we've talked about using AI as an advisory

editor for revision.

But there's still one last massive hurdle before this stuff goes to the user.

The dreaded QA phase.

Yes.

Quality assurance.

The last line of defense between our flawed documentation and a very frustrated user.

And this is Ghetto's third workflow.

And honestly, this is where AI really shines because it excels at repetitive rule-based

tasks.

The stuff humans hate doing.

The stuff we are terrible at.

The AI acts as this tireless proofreader.

It can scan massive documentation sets in seconds for inconsistent product names, broken

cross references, missing alt text on images, or even tone mismatches, right?

Exactly.

But it sounds like a spell checker on steroids.

And my worry is that people will just assume automation equals accuracy.

What about false positives?

It might catch an obvious error, but it doesn't really understand the context.

You're totally right.

It often misses subtle contextual errors or mismatched version info that is actually intentional

for a legacy product.

Right.

Because it doesn't know the history of the software.

Exactly.

So the optimal method here is to let the AI do the first pass to just generate a report

of potential inconsistencies.

A diagnostic report.

Yes.

Humans can even use shared AI dashboards to track these things over time.

Like if the same terminology issue keeps popping up, you know you have a systemic problem.

Again, human oversight remains completely mandatory.

The human reviews the report and decides what's actually an error.

So the AI points the flashlight, but you still have to decide what you're looking at.

Perfect way to put it.

But none of these three workflows drafting revision QA, none of the matter if the whole

system is just built on a totally irresponsible foundation, which brings us to the most important

part of Ghetto's article.

He outlines four cross workflow principles, essentially guard rails to keep AI from just

becoming another source of noise.

Right.

And what's the first one?

Number one is clarity.

Because as we discussed, perfect grammar means absolutely nothing if it confuses the reader.

Yeah.

If an elegant sense doesn't help the user solve their problem, it's a useless sentence.

Exactly.

Number two is accuracy.

AI is a predictive text engine, not a database of facts.

It cannot verify facts like a real subject matter expert.

It hallucinates.

It guesses.

It guesses very confidently.

So human validation is non-negotiable.

Okay.

What's number three?

Accountability.

And this is huge.

Humans are always legally and ethically responsible for the published content.

You can't just blame the algorithm if a server crashes because of a bad instruction.

No, you cannot.

You have to worry about data, privacy, bias, compliance.

The organization owns the output.

And finally, number four is transparency.

Which means what?

Practically.

It means sharing your AI practices openly to build trust with your stakeholders.

Don't hide the fact that you're using these tools.

Be clear about where and how they are deployed.

So people know there's a human at the wheel.

Exactly.

To sum it all up, the core thesis here is really encouraging.

AI won't replace you, but it will fundamentally reshape your time.

By offloading all these mechanical tasks, the scaffolding, the tedious QA tech writers

can actually focus on what matters, solving real user problems and designing much better

information systems.

It elevates the role entirely.

But I want to leave you the listener with a final thought to mull over today.

If AI eventually handles all the repetitive scaffolding, all the checking, all the formatting,

how will your core identity as a technical writer evolve?

That's a great question.

Do you become less of a traditional writer and maybe more of an information architect or

maybe a dedicated user advocate?

It's something we're all going to have to figure out very soon.