The Harsh Truth About AI in Business (You're Not Ready)

"Today is the dumbest AI will ever be.”

That’s what Johnson Goh, CEO of Mindhive, believes, and he’s proving it by quietly building the unsexy AI that actually grows businesses.

Forget the hype. Forget the flashy AI demos.

This episode dives into how AI is solving the boring-but-critical problems. Accounting, logistics, and operations that actually scale companies

In this episode of Tech It or Leave It, we cut through the AI hype and talk about :

✅ The biggest mistakes founders make when “AI-ing” their business
✅ How to build AI tools for teams who don’t even understand the tech
✅ Why back-office bottlenecks (not flashy AI demos) are where the real ROI lies

Cut through the noise. Trust the guy who’s actually scaling businesses with AI. This is the one you need to hear.

Transcript

Today is the stupidest that AI is going to be. Even though it's cool, it makes no sense to look into it. Until the androids come in, no one truly can predict the trajectory of AI. How do you forecast the next day what's going to sell? AI would definitely kill off some jobs. I don't think it's a matter of whether we're ready or not. I don't think we have a choice.

Welcome to Tech It or Leave It, a podcast for entrepreneurs who want to build smart, scalable tech without needing a tech degree. After eight years of building million-dollar apps for founders from around the world, I've seen what works, what flops, and what founders always get wrong. On this show, I break it all down: the real talk, the dumb mistakes, the moves that actually grow your business, from product ideas and development to automation, AI, and real business outcomes. Whether you're looking to streamline your ops, turn your idea into a SaaS, or just stop feeling left out of the tech conversation, you are in the right place. Let's make tech make sense. This is Tech It or Leave It.

Welcome to the first episode of Tech It or Leave It. Our guest today isn't just talking about AI. He's quietly building it into real businesses while everyone chases the hype. I'm talking to Johnson Goh, co-founder and CEO of Mindhive, the AI startup powering operations for brands like Zus Coffee, one of the biggest coffee chains in Malaysia. In just 12 months, Johnson and his team have gone from a scrappy startup to a top 13 finalist in the 2024 Alliance Bank BizSmart Challenge, all by solving the unsexy but essential problem with AI. We're talking logistics, accounting, and ops bottlenecks. Not flashy, but these are the tools that keep businesses moving.

Before this, Johnson launched a resale platform called Unearth. But like any sharp entrepreneur, he saw a better opportunity and pivoted fast. In this episode, you will hear the biggest mistakes founders make when trying to AI their business, how to build tech for teams who don't even understand the tech, and what it takes to really stay relevant when AI evolves every single week.

Welcome to the show, Johnson.

Hello, Adrian.

Alright, cool. That was a quick intro. How was it?

Good.

Good, good. Alright, cool. Let's dive straight into one of our questions today, which is something I'm very interested in as well—which is: what made you jump into AI so early on, that you are essentially one of the early AI startups in Malaysia?

So this stemmed from me running my previous company called Unearth. It's a very manual business and there was a lot of coordination needed. If you fail one single point, it falls apart. You can lose money just from forgetting to do something. So this made me constantly think about automation.

I remember during that time, ChatGPT wasn't out yet. Something that was starting to gain a little bit of popularity was those article writers. I don't remember the name… oh yeah, it's called Jasper, Jasper.ai. I started playing around with it, and instead of asking it to generate articles, I asked it different kinds of questions. It wasn't very good at that time, but it was kind of good enough. That triggered me to think, hey, this thing is going to just get better from here.

So I then figured out what needed to be in place to start an AI company. That boiled down to three things. First, a good tech co-founder. Luckily, my CTO at Unearth was a really good tech guy, but the problem was he was stuck in old technologies. I had to slowly convince him, give him examples, and when ChatGPT came out, I got him to try it—start generating Python code—and he was impressed. He dove into it, and I finally got him to join us.

Second, we needed a sandbox to play in. I was lucky because the founder of Zus Coffee was an investor in Unearth. Through our conversations, he mentioned they had a lot of use cases but couldn't find people to build them. That gave us the sandbox to play and learn.

And third, of course, capital. In the beginning, we funded it through jobs we took from Zus Coffee, but later we raised a small round of investment to grow the team even more.

A lot depends on whether technology is core to your business. If you run a gym business and just want a tech component, having a CTO may or may not be necessary. I can only speak from my experience. The way I attracted CTOs or technical partners was by first proving business value. I would build the whole operations of the business via Google Sheets or no-code tools like Airtable, generate revenue, and then approach tech people I knew—keeping them updated as the business grew. From there, I’d offer them to join.

With Mindhive, this wasn’t possible because it’s a very core tech business. That’s where we step in—to help others build the solutions they need.

Through networking, I built a relationship with the CEO of Zus. When I started Unearth, I reached out to him to invest—and he did. I kept updating investors, built trust, and we became very close friends.

Yes, they were already running Zus Coffee then. Early days though. Maybe 80 outlets, which was big but not like now—it’s 10x bigger.

Financially, we haven’t been in a tough position. But early on, AI was very nascent. Not much good material online, so we had to figure everything out from scratch. Many times we felt like giving up, but we pushed through, talked to people, and found solutions.

What kept me going was the belief that today is the stupidest AI is going to be. Improvement will accelerate. It’s not a matter of whether it can, but when it can. So it’s about managing the timeline so we don’t die halfway.

AI is an enabler. Don’t just plug tools in. First, understand your business and map your value delivery process. Identify the gaps that cause problems or leakages. Start with the problem, not the solution. Then solution from there.

Too many business owners go for shiny objects without ROI. That’s the biggest waste of time—trying tools without clarity on problem statements. It’s not plug-and-play. Change management and integration take time. Invisible wastes of time pile up.

We helped Zus Coffee automate their accounts payable and accounts receivable processes. With AI, invoices are dumped into a folder, automatically classified to the right outlet, and synced into the finance system.

We also worked with a customs broker firm, automating their end-to-end workflow for customs clearance. That saved them 70–80% manual manpower, solved scaling issues, and reduced reliance on repetitive human labor.

For forecasting and simulation, we helped Zus Coffee outlet managers predict weekly hourly sales and forecast what food would sell the next day. Each outlet and sales channel needed its own model because factors like weather, holidays, and promotions affected them differently. Experimentation with data features was key—though we eliminated weather due to poor-quality data in Malaysia.

The short answer to “is there a business not suitable for AI?” is no. Every business can adopt AI in some way. The question is: to what extent? For physical businesses, AI might only handle back-office tasks like order processing or accounting. For others, it enables scaling and revenue growth.

AI will definitely kill off some jobs. Businesses adopt automation when they can’t find people for a job or can’t afford them. But humans always find the next thing to do—just like we don’t have telephone line connectors today. The shift will be in the type of jobs we do.

To stay relevant, the key skills are having vision and asking good questions. Hard skills may decline in value, while thinking and problem-framing will matter more.

In Malaysia, I don’t think it’s a matter of readiness—we don’t have a choice. To adopt AI effectively, companies need to start practicing good data management. Clean data means collecting it correctly, holistically, and integrating it into workflows. Missing data points, like untracked out-of-stock events, distort AI training and create inaccurate models.

The premise behind Mindhive was starting as a service company to learn, build competence, and grow a strong team. Zus Coffee gave us use cases, data, and feedback, which helped us develop. Over time, we are building products, like OrderMaya, to help wholesalers, manufacturers, and distributors deal with the shift toward smaller and smaller orders. AI will coordinate processes that are otherwise difficult and costly to scale with human labor.

In five years, we see Mindhive as both a service company and a product studio—with a portfolio of five, ten, maybe twenty products, alongside a strong services arm functioning as R&D.

If I weren’t building Mindhive, I’d probably be building another company.

You can find more at www.mindhive.asia or our product at www.ordermaya.com. You can also reach me on LinkedIn, Facebook, or WhatsApp.

Proud to be the first guest here. Thanks for having me.