Building Dorje and what VCs get wrong about AI in emerging markets
The former AWS head of AI and ML for ASEAN bemoans widespread ignorance about AI hype among VCs
What does selling artificial intelligence (AI) solutions to ASEAN enterprises, building companies and successful Asian podcasting have in common? One person -- Bernard Leong.
Bernard, or *Doctor* Bernard Leong, as he cheekily emphasises, is the irritatingly multi-hyphenated operator/investor who has dipped his fingers in pretty much everything — from startup building to startup failing, sitting at the forefront of digital transformation at companies like Airbus, Woh Hup, SingPost, and selling AI to enterprises for Amazon Web Services (AWS) long before Gen AI was even a thing.
Oh, and he also runs a 113,000 follower strong podcast called Analyse Asia.
Today he is the co-founder and CEO of Dorje.AI, an all-in-one platform which automates financial workflows for companies like checking and auditing documents, invoicing, appending, approvals and so on.
He proudly states that Dorje has only raised $250,000 to date and if all goes well, this will be all the money he needs to raise, ever.
An avid technologist, Leong is neck deep in the day-to-day process of business building. While everyone else is barely grasping the wonderous and horrifying potential of Gen AI, he is already out there riding the bull run.
It’s because of this that makes him highly cognisant of how AI is dramatically transforming startup business models, fundraising requirements and tech teams. You can’t remain stagnant, he says.
“I’ve been programming since the age of nine and even my programming approach has evolved significantly during this time. Every five years, there’s a new programming paradigm. But so many people continue to think and operate in a box,” opined Leong on The Upside Podcast.
One time, his team spent 5 months trying unsuccessfully to code a product feature. He completed it by himself on a Kuala Lumpur to Singapore-bound flight (total coding time was 20 minutes, he says) after punching in a couple of prompts on Cursor. It worked beautifully. He fired the entire team.
“A lot of big development houses are going to start suffering because the cost for a normal person to build today is so low,” he added. His latest venture, Dorje.AI, is itself just a 3-man team and he plans to keep it lean.
He bemoaned the lack of hands-on knowledge in the VC ecosystem which is leading to misunderstandings about how best to value or think about AI in today’s world. Among his other “contrarian” views is that AI foundational models aren’t as commoditised as most investors like to believe.
“I always ask my customers in AI — Are you using AI to do things better or to do better things? They’re very different. Doing things better is about cost efficiencies. Doing better things is about changing your business/operating model. LLMs are getting better today, which means it’s allowing you to do better things…If semiconductors are as commoditised as these bankers say, why is TSMC the ninth most valuable company in the world?” asks Leong.
“(So) semiconductors are still improving. Moore’s law is still giving you an exponential rate with AI. It’s just that now you have AI versions of Moore’s law, so it’s a never ending competition. So there’s no true full commoditisation whether it’s a semiconductor or LLM,” he explained.
In other words, it’s early days yet for many AI startups in Southeast Asia and globally, most of whom are still in the stages of finding product market fit, like Dorje.AI. Using subscription SaaS models in AI is also unlikely to work in emerging markets, because the costs of labour in these economies are simply too low, he added.
All these are important considerations which need to be baked into the business model and operations of AI startups across emerging markets like Southeast Asia and even the Middle East.
“A lot of VCs from our part of the world have not done actual work in selling enterprise cloud solutions,” he said.
“(But) I’ve moved AWS’s business from six to eight digits in two years. Trust me, it is extremely difficult to do. Yet one Southeast Asian VC once told me that this was something that could be done by two kids from university,” laughed Leong, who led AWS’s AI and machine learning team for ASEAN from 2019 to 2021.
His goal for Dorje.AI? To build a self-sustaining business and never raise from VCs ever again. The journey is certainly early and he’s hoping to have the last laugh. Time will tell, but he’s quietly confident that he will.