{{big-quote}}

‍

AI can translate your IPO or ESG report—but can it do it right? Colin Hong, Co-Founder of DeepTranslate, reveals how his Hong Kong startup outperforms Google Translate and ChatGPT in the most high-stakes corner of translation: finance and legal compliance.

In this episode, we cover:

• Why niche AI is the only way to beat Big Tech
• The hidden 20% translation gap that LLMs miss
• How data compliance fears hold back corporate AI adoption
• Cashflow discipline and hard lessons from startup life
• Building a team with two PhD co-founders (and why the “sales guy” matters)

‍

Transcript

Jonathan Nguyen (00:00)
What does it take to walk away from a 25-year corporate career running divisions in Hong Kong, Beijing, Shanghai and Chengdu, managing hundreds of people and working with clients like HSBC to start a tiny AI company that dares to take on the giants like Google and Microsoft? Colin Hong is the co-founder of DeepTranslate, a Hong Kong startup translating high-stakes IPOs, ESG reports and legal documents between English and traditional Chinese. Drawing on decades in the financial printing industry,

Colin's team has built a niche AI that delivers higher quality translations and helps clients cut turnaround times in half. In this episode, we unpack why going niche is the only way to outplay big tech when you can't outspend them and why partnerships, cashflow discipline, are the difference between surviving and shutting down.

Welcome back to another episode of the Unsensible Podcast, where we speak to the unrelenting founders on their quest to change the world. And today, one of those founders is trying to change the way that governments and large enterprise translates one of the most important languages in the world, which is traditional Chinese. Colin, great to have you on the show.

‍

Colin Hong (01:23)
Thank you for inviting me.

‍

Jonathan Nguyen (01:26)
I'm going to ask you to pitch because this is the first time on the podcast and a pitch 30 seconds is going to be the best way for us to get to know you.

‍

Colin Hong (01:36)
My name is Colin. I'm the co-founder of DeepTranslate. What we do is using AI to do translation for compliance documents, especially on financial documents like the annual report, ESG report, IPOs. We do a lot better than any traditional AI translation tools like Google's or any other that you may be using now. We double the qualities, the accuracy of the translation, and we help our clients to save 50 % of the time to do the traditional translation work. So that's basically what we are doing now.

‍

Jonathan Nguyen (02:05)
Probably the one of the Christmas pictures I've had so far. ⁓ I think I can understand why Colin I've been doing a little bit of light stalking. I'll call it research. And I can see that you've had quite a lot of experience in corporate before you started to translate. Can you tell me a little bit more about that?

‍

Colin Hong (02:24)
Before I found DeepTranslate in 2018, my job experience was in financial printing. What financial printing industry in Hong Kong is doing is they get up a listed company to do the typesetting, translation, and filing to different stock exchange. I have been doing this line of work for over 25 years. I was working in a listed companies in Hong Kong, which is the largest financial PR companies, which help offer

400 listed companies to do their financial PLL, role shows and financial printings. And I'm in charge of the financial printing divisions in Hong Kong and Beijing, Shanghai and Chengdu. I had experience to do what is called ⁓ BPO a business professional process outsource on printings for large corporate. Used to help HSBC, one of our largest clients at that time, on all the print...procurement in APAC. So I have been traveling a lot to China, to India, Australia, Japan, and Taiwan as well.

‍

Jonathan Nguyen (03:29)
So really deep subject matter expertise. me a bit more about your co-founders. You've got a couple of PhDs on deck as your co-founders. that right? ⁓

‍

Colin Hong (03:38)
The reasons why I co-founded this company is, know, when you are in a corporate, you usually go to events and then you meet a lot of people, you get a bunch of business cards. Usually when you go back, either you have time to scan it or you just put it on a desk for a while and then you forgot who they are. So I actually went to one of those events and after the event, two weeks later, I got a call from someone I don't even remember. And then she said, well,

I met you in a certain event and I have a friend who's a professor in Hong Kong universities. He was doing a study in AI and then I'd to link up with you. And I am a big fan of technology. So I met Professor Chien, he's one of our co-founders now. He was in Hang Seng University helping them to start a new AI department. One of the projects is AI translation on IPOs. And I am actually in charge of the financial printing.

divisions at the time in the listed company. So I'm very interested to work with the professor. So nowadays universities not only get the fundings, they have to make sure that funding that research is actually can be used in the commercial world. So we work together as I provide information inside of the AI translation in 2017. We work for a year and then we decide to start making commercials.

commercialise these products. At first, we started to discuss with the universities and corporate. It's a lot of issues and then it didn't work out and all of a sudden, Professor Shin told me, hey, why don't we do it ourselves? We invite another PhD and other founders to join this team. And then we started a company in 2008.

‍

Jonathan Nguyen (05:24)
That's amazing. It was kind of serendipitous in a way because you bumped into someone. So you've got Professor Francis Chin. What's his specialty? ⁓

‍

Colin Hong (05:33)
Professor Francis Chin is, he used to be a professor at University. He's an AI expert recognized by the Stanford universities as the top 2 % scientist. So he has a research paper that over 20 years and still been using by a lot of researchers. That's what he's good at. And then I have another partner, Dr. Bethany Chan. She's the most talented lady that I ever met. She is a PhD in computer science.

So she's very good in tech. She used to be a barrister as well. So she got a law degree and she also a CFA as well. So she's very good in finance. Most difficult exams she got all got those exams all passed and.

‍

Jonathan Nguyen (06:17)
She's like the perfect, perfect Asian parents child. She's got the finance degree, the law degree, and she's also a computer scientist.

‍

Colin Hong (06:22)
Ha Yes,she started teaching in Chinese universities when she was 18. So she is a genius.

‍

Jonathan Nguyen (06:36)
Okay, so I guess these are the scientists and then you're the sales guy, right? You're in charge of growing. ⁓

‍

Colin Hong (06:42)
keep telling people I'm the non-tech guy in a tech company so I don't know anything about tech and then I suck and then I company

‍

Jonathan Nguyen (06:49)
Well, every company needs that growth person. Otherwise, you know, it's very good tech with absolutely no revenue.

‍

Colin Hong (06:56)
Exactly. See, the only talents that I have compared to my partner is that they may be very talented in many, many ways, but they are not very good with people and they are not very good with selling. I guess you met a lot of IT person as well and I love startups. So when you speak to those professors and founders, their babies know that these startups...

They see a lot of values, but when you negotiate with your clients, there's lot of tactics you need to give some in order to fix that, right? I think companies need have those persons to get deals. Otherwise, everything is just on paper.

‍

Jonathan Nguyen (07:34)
Our business exists because of these very smart people. Basically, I feel like I have two ears for a reason, right? So one ear is you listened and get really interested by the science and get excited by it. And then the other ear is you're listening to what normal people talk about and think about, and it's definitely not the way they see the world. So being able to pull in those two together has built us a business basically. So I totally understand you. Tell me about how you got into and why the tools don't work so well. for local market if you're dealing with chat GPT, Gemini, Llama, all these

‍

Colin Hong (08:08)
First of all, we are a very niche product. We focus on the financial market, legal documents, Chinese to English or English to Chinese translations. When I first started in 2008, actually no one really know what AI is. But Google AI translation exists. A lot of people have been using that tool for years already. We are only a startup.

We are smaller than SME. We are a tiny, tiny company. How could I beat those Google's and Microsoft? We have to do something very niche and that part of it, we do better than anyone else. So we focus on financial documents. And since I have over 25 years experience in the financial documents, I know the pain point. I know what the user needs. I know what the clients want. So

I focus on that and provide a very high quality translation. I believe with all this LLMs nowadays, people will look for very niche products because we cannot compete, to be honest. I don't think any startup can be competing with Alibaba, Mattes, Microsoft or Google. They burn the money to get their market, but we're not that. So we provide various niche solutions and I believe that will be a trend for most of the startups on our business. Supply is very important.

Imagine that you have M &A documents, billions and billions of dollars, you put on GPT or any other free models. That's a big risk there. We always say that whether you are buying your product or you are being a product yourself, right? When you buy a product, you pay and get some product you can use. When you use a free product online, you basically other products, your data will be uploaded to the server and then useful many, many different products.

So in Hong Kong, there's research that says the adoption of AI in financial market is over 80%. It's a very high numbers. But then when we speak to clients, they all want to use AI, that's for sure. But not every single one is willing to pay. There's a lot of reasons because of the data security, because they don't know it, because they are afraid that AI will take away their jobs. Most people in corporate world is that, well, let's see what are my competitors using.

any AIs before I move. I don't want to be the first, right? I don't want to be the guinea pig. I want to see how others do. But then that's strange because the research tell you that over 80 % of people are using AI already in the financial world. So how come? That's because a lot of people is bringing their AI, bring their own AI. They bring their free tubes or whatever tubes, which is non-compliance to company, but no one cares because I don't want to.

‍

Colin Hong (10:59)
First, everyone is doing it and I don't want to spend two days to draft a proposal anymore.

Jonathan Nguyen (11:04)
Let's dig into that. said you're a very hyper-focused niche. What are some examples in financial documents that regular LLMs always mess up? ⁓

‍

Colin Hong (11:14)
Actually, LLM has a lot of problems. First of all, the consistency is very bad. If you translate a single sentence 10 different times, it gives you 10 different results. There's also a hallucination of AI, everyone knows that. Let's say when you translate a document, put down Mr. Chen in that document, it may give you a timeout ball, but that Mr. Chen may not be the correct translation.

You just make up something and then it looks really real. A lot of people doesn't really know that when you do a translation, sometimes when the LLM tried to give a very fluent translation, when you look at it, you understand it, but they did not translate word by word. There could be over 20 % of the content will be gone. So imagine there's a legal document. Lawyers will argue on a single board for two days and then now you have 20 % of the content missing. That's

could be a very big problem. Researches came out this year is that LLMs is not very good in translation number as well. So that could be another problem. We find that due to political reasons, sometimes they won't handle certain contents on certain models. So that could be another problem as well. So imagine that you're doing an IPO yourself and then you are in a very new industry that I never

see we don't have any data in any sort of change before. So this transition will be very bad because we don't have the data. But nowadays we have the error layer. So this help us a lot in those content as well. So I will say we're really competitors against those error models. We help each other.

‍

Jonathan Nguyen (13:01)
Yeah, so your biggest competitor is probably a person.

‍

Colin Hong (13:04)
Yes. Believe it or not, a lot of people still don't want to use AI. And I met some translation firm, we discussed, we showed them, we give them the free trial account and they never use it because they actually afraid AI will take away their jobs. But I always believe if you don't embrace AI or any technologies, then it's not that AI will take away your job. It's the people who using AI will take away your jobs. So yeah, it's just...

And on Tuesday we have...

‍

Jonathan Nguyen (13:35)
In the history of time and commerce, I've never seen a company say, well, you know, because we've got this new tool, we can process things 20 times faster. We're going to lay off 20 times more people and then stay where we are. No one's ever said that everyone said. This thing's going to make us 20 times faster. Well, let's make 20 times more. Right. Yeah. That's how things play out. So I think adaptation at an employee level is super important. I for, for our team, especially people say.

Marketing is going to, AI is going to take away marketing jobs. It's going to take away marketing, like grunt jobs, the jobs that people didn't like anyway. You know, running those reports, buying media, programmatic buying of media, all of the generating variations. This type of stuff is so like manual and people, if that was the one thing you're good at, well, you're not going to want, you know, an AI to take it away, right? Do you have plans to scale out within Asia?

‍

Colin Hong (14:33)
at the moment to be fair. Yeah, that's not first of all the overall economies around the world I think is still not very stable. So for SME, we have to be very careful on every step we take. I want to be Alibaba is able to invest a lot of this and that and see whether it went well, but we cannot do that in ourselves, a small company cash flow is everything.

you need to maintain a very good cash flow. If you make a wrong step, you won't see me next year. So that's, that's, think very important. It's actually not my first time to start a business. I started my first business when I was 30 years old. And then even though I have clients at the time, they do very good ⁓ sales, but then I did not do well with my cash flow. And then one day I don't have all this funding. I don't have the support.

And then I don't know what to do the next day. And then eventually I went on business at the time. So cash flow is very important. So currently I don't have a plan to do outside China, Hong Kong or Taiwan.

‍

Jonathan Nguyen (15:42)
Yeah, I think a lot of founders discovered the hard way in 2021 and 2022. Yeah. know, the importance of cashflow and how much quantitative easing really helped their business. Yeah. So you've got, I saw some announcements. had an, you made an announcement about carbon exchange last year. Walk us through like how this, these types of partnerships are like beneficial to, you know, tech startups like you.

‍

Colin Hong (16:11)
I think, see as a startup, I mentioned I wanted to do Cantonese. Actually, I want to do a lot of things in the bank. But then as a startup, we cannot do a lot of things because we don't have the resources. Not only we don't have the money, we don't have the people do it. So basically it's difficult. And partnerships are important. Since those carbon exchange or other partners, are doing the niche market, the niche products, but we are all...

have the same clients help. So basically we all went to, we're looking for financial institutions or listed companies, all our target clients. So why not work together and then get a bigger market? then when the client does not have a need for translation, they may have need for ESG report and then pass the clients to them so that at least when I go to my clients, imagine

Every time I meet you, just say, hey, how's it Do you want to use some translation? Yeah, translation. And you won't talk to me anymore. I could introduce you. that's my partner. He has a, a, a report that help you to build your, a years report in a very, in a shipper way and fast, which is compliance to the Companies ordinance And this is my, partner who is, who is a financial PR.

It helps a lot of list companies to do listing and broadcasting nowadays. So I think forming partners is very important for startup to survive as well.

‍

Jonathan Nguyen (17:49)
Yeah, not just startups, actually. Some of the biggest tech companies in the world when I was, when I first started out, well, first of all, I started on the tech side and I dealt with the different layers of tech companies. Right? So sometimes, you know, you buy stuff from Microsoft, you don't really buy from Microsoft, but you buy it from their channel partners. You buy it from, you know, service providers or consultants who sell through. So on that side, yeah, I dealt a lot with that. But then when I went into marketing properly, I dealt with them, Microsoft, IBM, the big tech players.

And then you see actually the levels of partnerships and how many thousands of partners that they have and actually the power of having that ecosystem, the leverage it gives them. So that no matter who you talk to, they're going to sell you one of three vendors anyway, right? Like it's good to see that, you know, you guys, you guys are doing it and yeah, hopefully we'll, we'll see some more announcements from you soon. Talk a little bit more about, you know, moving, went from corporate to startup. Were there any habits that you had to unlearn? ⁓

‍

Colin Hong (18:47)
think tons of it. Basically, for anyone who moves from corporate to a startup, first of all, my suggestion is think twice before you do it. See, when I used to be in the corporate, I didn't even know I have so many resources. I had asked people to do this and that, and then they give me a report and then I go to my clients. And then now I am the person who did this and that and then still deliver the report to my clients. I always say that.

Now in corporate, you focus on managing people. You focus on managing your boss expectations. You focus on the board. You focus on how to deal with your employee as well. Basically, I'm a link to everyone. But in a startup, always say, what I am in charge is payroll to payroll, right?

So basically you do everything. You go to supermarket to buy those stuff and then give it to your staff. But then you also go to the bank to pay the salaries. And then you are both the account receivable and account payables and the HR, everything. So actually you have to really unlearn a lot of things. And then also when you go to a, when you from a corporate, the corporate I used to be is

big companies with over a couple hundred people and they got a very good brand and also a lot of listed companies. It's a lot easier to make appointments when you're carrying a big corporate brand. They recognize your brand, right? So even though I know the same person and I go to them with SME or a startup, situation change. All of a sudden they don't have time for me. So that's a different...

working with a big corporate and a startup. But then, course, in a corporate, every single decision is difficult. You have to deal with a lot of stakeholders. You have to deal with a lot of peers. Right? I always said in a big corporate, that when we have monthly or weekly meetings in a big corporate, right? But those meetings is not important. Those meetings are very important.

You have to go to different pre-meetings to deal with different stakeholders to get them agree on certain things before you go to the meetings. The meeting is just a format, put it this way. So that's very important. But in startup, you have to be very flexible with your clients, your peers, with your employees. I no longer can, you know, when my employees come back at 10.

I won't have an HR go to them and send them the morning that's in, right? So in the startups, you have to be very flexible. But then in a corporate, when you're over 100 people, you give flexibilities to them, it will be a big panic for yourself. It will be a chaos, right? And then people will complain and then once they're coming back at 10, everyone will be coming back at 10. And then...

And then there was a lot of race, and then someone will be 10, 30 and be awesome, right? So you have to really, you have to learn a lot of things and learn a lot of new skills.

‍

Jonathan Nguyen (22:05)
What you said about, know, when you were in a corporate, you could get a meeting with anyone fairly easily. And then you switch to a small company and then it's harder to get a meeting. You're the same person. And I always tell people, this is like your lesson, your first lesson in branding, because ⁓ nothing is different. You're the same. Anything that's different is the name on your business card and what it implies, right? The weight that it carries. mean, people, especially founders don't really think enough about it. They think.

that, I've got a really great product. So if I sit in front of you, I stand in front of you and tell you about my product, you should want to buy it. You should want to hear about it. Right. And it's not the case, you know, when you're talking to me about your product, I'm thinking about, know, can I get home in time to cook dinner? Like where are my kids? Is the cat okay? Like,

‍

Colin Hong (22:52)
True, true. On every single event or speech, actually people only take away one or two things, right? So even you have a one hour presentations or an hour of this podcast, after a day, they may only remember one or two things. So that one or two things is very important to them. And then also branding, as you mentioned, is very important as well. So as a startup, I actually, used to only wear suit ties and all that, right? Now as I got my long hair,

gray hair. To be frank, it not have a very professional appearance. To me, I think it is a very good brand. When you went to people, you went to an event, there were hundreds and thousands of people in that event and then everyone looked the same. Everyone with glasses, with a suit, with black suit, white shirt, maybe with a tie nowadays, then with glasses. Everyone looked the same. Short hair, skin hair.

on the side, everyone of the same. So you need to do something to let the people remember you. So that's one thing that I do. I believe that that helps as well.

‍

Jonathan Nguyen (24:03)
You definitely look more like tech now than you do a banker. So I remember, I remember I messaged you when I saw one of your headshots media once you were in a suit, you were very buttoned down, very serious. I was like, that doesn't look like the guy I know. That doesn't look like the tech guy I know. So we're coming up to the hour and I want to leave everyone with two things really. Number one is what tech do you use in your daily work life that you can't live without?

‍

Colin Hong (24:29)
Um, I will say AI translation. Basically, it is AI translation is really one of the tag that I use because, um, it's very important, um, uh, to write in different language nowadays to different people. That's very important. Another thing is definitely the LLMs. So, uh, when you draft a proposal, when you draft emails, now see people use, it's not

‍

Jonathan Nguyen (24:31)
Ha ha.

‍

Colin Hong (24:58)
Too long ago, people was using Grammarly to help them to write. A lot of people are still using it. But with LLM nowadays, you don't really need Grammarly to be frank. And then actually I've rolled a message and then use the LLM to proofread and help us improve it. So that's something that I use a lot. And then I have been dealing with foreigners and then also mainland Chinese and Taiwanese. So I have different clients.

To write the Chinese in a certain way, definitely need AI to help me as well. And then I use LLM to do another things. It's helped me to do brainstorming. So I think brainstorming is very important. We used to have a team of people sit down to give different ideas so that all of a sudden you may get new ideas and then...

to do that. I, nowadays I use sometimes to help. ask LLM to be my CFOs, my base CTOs to be my design cooperatives. And then we discuss. even though I'm discussing it with myself, but we discussed so that, and I find it very, very useful that when you ask your AI to ask you a question, it actually helps you thinking. And then it's actually generate some ideas that I never thought of. So I think that's very, very useful. So basically.

is LLM and translation. this most used to the tubes.

Jonathan Nguyen (26:24)
I would say that for me, I do use those two things a lot as well. Yeah. Totally understand that. In the next 30 seconds, I'm going to ask you for the answer that comes to the top of your mind right away. Test book or resource that you're reading right now.

‍

Colin Hong (26:38)
I actually go to a lot of YouTube channels to learn about cryptocurrency and I use a double speed to learn it. I think crypto is for me. Yeah, you too.

‍

Jonathan Nguyen (26:50)
YouTube. Okay. So if you weren't a founder, what would you be doing instead?

‍

Colin Hong (26:54)
I I want to travel around the world. I want to see the world.

‍

Jonathan Nguyen (26:59)
One piece of advice you'd give to a founder who wants to start something in Hong Kong or the greater Bay area. Don't do it. And what is, do you have a prediction for AI in the financial services in the next two years?

‍

Colin Hong (27:05)
Thank you guys.

I think the adaptions will be close to 100%.

‍

Jonathan Nguyen (27:17)
Cool. Big call. think I'm with you. Okay. Thank you very much for joining us on the podcast today. And I want you to leave us with one exciting development that you've got coming up.

‍

Colin Hong (27:28)
We are currently developing a tool that help audit firms or anyone who need to check against the documents on all these numbers. So basically what I mean is when you have a financial documents, say, and then it reports a bunch of numbers like the top line, bottom line, very last years and all that. Within the text, there's also numbers referring to those tables as well. So we are actually building a tool that help people to

scan the documents to identify whether there's problems and issues in that. We found that in a lot of first documents actually there's a lot of mistake, even the proofread by a lot of professionals and they still make a lot of mistake and post it online and file it into the stock exchange I think this is actually helped a lot of people to save a lot of time to do their current jobs. But we'll see what happens.

‍

Jonathan Nguyen (28:22)
Thank you so much for joining the show and I am looking forward to seeing you again. ⁓

‍

Colin Hong (28:26)
Yeah, thank you.