Using AI Beyond Marginal Gains

I've been using AI for quite a while now. It has definitely helped a lot in driving efficiencies particularly in the creative aspect of my work. But I recently started asking myself, "is there something more?"

As with many of you have noticed, generative AI tools like ChatGPT, Claude, and Gemini really shines in creative writing:

  • paraphrasing words or sentences
  • writing documents, reports, social media posts
  • creating plans and spreadsheets

If you go a step further than that, you can use these tools for further research and analysis. Personally, I prefer Perplexity for research purposes as the results are more accurate and recent. While the popular tools all now have web seach functionality, the default settings for the average user yields better results in Perplexity.

When it comes to the analysis part, I'd say they are relatively good depending on the model and prompts you use. In my case, I've used them for the following:

  • analyzing CSV files for keyword research
  • analyzing CSV and PDFs for reporting purposes (think Google Analytics and Google Search Console data exports)
  • analyze code and provide fixes /recommendations (think HTML, CSS, python, etc.)

Then there are your advanced users. These are people who developed complex prompts and delved (yes I used this hated word) in developing advanced system prompts or custom GPTs or AI agents. With this advanced use, it's easy to see 10-20% improvements in work/task completions.

I consider myself to be in this group.

But I've been wondering– is there more?


50+% Efficiency Gains

When I started researching about this, I had a belief of gaining over 90+% gains. But based on what I've been reading, that is mostly refuted.

What remains though are the unicorns that truly transform their work and business, generating over 50+% improvements across the organization.

I think that is where we are still heading. That for me is the allure of agentic AI.

But as that relies complicated setups and not readily accessible to the regular user, the question remains–how can we use AI today to transform (my work/my business/my organization)?

According to McKinsey, the promise of 50+% efficiency gains lies in a strategic approach, rather than a functional one.

Organizations that fundamentally rethink business processes rather than adding AI layers to existing workflows achieve dramatically superior outcomes. 

And that's where business leaders across all levels need to be aligned.

That's where managers need to learn how to manage and how to implement change initiatives.

The real culture needs to be addressed before this transformation can happen.

Sidenote: culture is often portrayed superficially as perks or disguised under the words or adjectives the C-Suite often says publicly. But real, deep culture goes beyond that. It's the underlying system of shared values, beliefs, attitudes, behaviors, and norms that shape how work is done and how people interact within an organization.

As you can see, no matter how much the CEO proclaims the culture they have or their AI initiatives, if this breaks down somewhere in the org chart, this transformation can at worst, easily become out of reach, or at best, delayed.

More to Explore

These are just some thoughts running through my head right now. Obviously there are a LOT of things I don't know or even grasp about genAI. And with the speed of development, it's hard to keep up.

One of the things I'm experimenting on is agentic AI, MCP servers and all those autonomous use cases for AI.

I'm not even diving deep into special AI use cases like image creation (dall-e, midjourney), videos (veo, invideo), voice / audio (descript, elevenlabs), and many more.

There's simply no time to learn and test all these tools.

Gen AI Is Creating a New Problem

Underneath all this generative AI hype, a new problem is emerging.

This problem is the oversupply of content and limited availability to filter and consume them.

Think about it. With the ease of creation, one can create thousands of pages in just a few hours.


Couple this with vibe coding, you can create multiple websites or social media posts and videos easily.

And I'm mot even considering the cybersecurity problem all these present.

This is what Google is actually trying to prevent–focusing on helpful content and their EEAT guidelines to filter out generic and regurgitated content.

But even then, that can be easily bypassed.

So what do you do now as the "end-consumer" of these content?

  • How do you filter out the noise?
  • How do you find real credible guides or brands to follow?
  • Who do you trust?

I don't have the answers.

All I know is that the way I consume content and the way I do my work have significantly change.

And I'm sure this is happening across the entire world.

Which means businesses need to adapt amd change the way they do things as well.

Ariel Lim

Ariel Lim

Management consultant / MBA / Inbound marketer who helps startups generate leads, create and execute strategies.

Leave a Comment

Make Your Competition Irrelevant

Enter your email and I’ll send you how to find hidden opportunities in your industry that everyone else is taking for granted.

Ready to grow your organic traffic?

Enter your email and I’ll send you the case study of how I grew my organic traffic by 110% in just 3 months!