The Last 10% Is Where AI Pays Off

"AI hasn't made quality less important. It has made quality dramatically cheaper to produce."

Everyone wants to talk about the big AI opportunities.

Agents running across the business. Complex automations. End-to-end workflows. Systems talking to systems. There is real value there.

But when we start working with a business, some of the fastest and most obvious wins are usually much more boring. PowerPoint presentations. Proposals. Board papers. Client reports. Internal documents.

The work every business already produces, every week, and spends a surprising amount of time making presentable.

That last bit matters.

For years, the final 10% of a document could easily consume half the effort.

You had the thinking. You had the numbers. You had the story.

Then somebody had to make it good.

Fix the structure. Clean up the formatting. Make twenty slides look like they belong in the same presentation. Turn a spreadsheet into a visual somebody can understand. Apply the brand properly. Get the spacing right. Rewrite the awkward bits. Make the proposal look like something worth what you were charging for it.

And for good reason. The final polish is often what makes the difference.

A good idea buried in a bad presentation is harder to understand. A strong proposal that looks thrown together carries less perceived value. A board paper full of dense tables makes the decision harder than it needs to be.

Presentation quality affects how easily people understand the work, and how they judge the work.

AI has changed the economics of that.

The expensive part just got cheap

I am seeing this increasingly as people move beyond simply asking AI questions and start using it properly inside their work.

A rough document can be turned into a polished, consistently formatted business document in minutes. A presentation can be rebuilt against a brand template without somebody spending an hour nudging boxes around PowerPoint.

The same company fonts, structure, logos, layouts and visual standards can be applied every time, regardless of who prepared the first draft.

A table of numbers can become a clear visual that makes the point obvious.

Work that previously needed the person in the business who was "good at PowerPoint", or sometimes an external designer, can increasingly be handled as a normal finishing step.

That does not make design or judgement unimportant. It does the opposite.

AI has not made quality less important. It has made quality cheaper to produce. And that creates a much bigger opportunity than simple time saving.

Quality no longer has to depend as heavily on who prepared the document, how much time they had, or whether somebody with design skills was available to clean it up.

You can start turning quality into a business standard rather than an individual skill.

The other side of the coin

There is a downside.

AI has made it incredibly easy to produce more stuff.

More emails. More reports. More slides. More words.

And we have all seen what happens when that goes wrong. Twenty pages where five would have done. Generic language. Presentations that look like AI produced them because nobody bothered to do anything with the first output.

More output is not the goal.

Poor AI use can make the problem worse. It lowers the cost of producing mediocre material, so you get more of it.

The opportunity is the opposite. Use AI to improve the standard while reducing the effort. Do the thinking. Check the substance. Then run the work through a consistent finishing process before it goes anywhere.

That final step is becoming almost ridiculously easy compared with what it used to take.

Look at the finishing layer

 

If you want to see where this sits in your own business, do not start by asking what you should automate.

Take one thing you already produce regularly. A proposal. A client report. A board paper. A presentation. Then look at what happens after the thinking is finished.

Who cleans it up? How long does that take? Who fixes the formatting? Who makes sure it is on brand? Who turns the numbers into something another person can understand quickly? Who takes the first draft and makes it look like something the business is happy to put its name on?

That is the layer I would look at first. Because the value is easy to see. You can put yesterday's version beside today's version. You can see the difference in quality. You can see the difference in consistency. And you can see how much time disappeared.

That’s important because early AI wins need to feel real. Not clever. Not theoretical. Real.

The last 10% used to be the painful part. Now it may be the easiest part to improve.

AI did not make the final polish less valuable. It made it dramatically easier to deliver every time.

Find out where your finishing layer is still costing you time it shouldn't

Most businesses can tell you what they produce. Few can tell you how much time disappears turning a rough draft into something they're happy to put their name on.

The 5 Minute AI Readiness Snapshot helps you see where that final 10% is still eating hours it no longer needs to.

Take the 5-Minute AI Readiness Snapshot

--

About the Author 

Mark Lucas is Co-founder and Managing Partner at Generation AI, where we help New Zealand business leaders build real AI capability.

Information provided is general in nature and does not constitute legal advice