My oldest sibling
Part 1: “You got nothing to do right? Do this for me la. (You cannot say no.)”
This story starts during my National Service days in 2024. My dad thought I had plenty of time on my hands. That wasn’t exactly true, but it wasn’t exactly false either. National Service is a weird period of time. So he got me to set up the e-commerce side of the business: listing products on our e-commerce website. (And yes, I didn’t get to say no. Fun times!)
I could have done it all myself, but I really didn’t want to spend my breaks (brainlessly) uploading products day after day. I was also studying for university entrance tests, so juggling everything was rough. Then I had a brilliant idea: what if I got part-timers to help? (Back then AI wasn’t that good yet, so humans were still relevant. Not anymore! Jkjk.)
So I got to work. I bet that the upfront cost in time would pay off: writing proper SOPs, setting up an Excel sheet to track progress, and teaching newcomers. For a while, it did! I had an army of 5 to 10 part-timers at any one time. The bottleneck became my dad, who sent me product lists less and less often.
This lasted about 6 months, until I left my trainer posting and hopped over to the Air Force.
Part 2: Ka kia for one week
Ka kia (Hokkien): “errand boy”. Basically a minion.
Exhibit A: a real conversation with my dad, sometime in April, after I came back from the UK
Pa: En Hao, now you’re done with army, and you have some time before your next internship, you should come down to help out at the shop.
Me: What you need me to do? Why suddenly need help?
Pa: Not enough people in office, all of them go overseas together.
(Context: back then, the entire sales team was made up of our relatives.)
Me: Can I even be of help?
Pa: You learn very fast, sure can one. Come during this one week in April can already, just to cover for their absences.
Me: But I want to relax, I never really rested… And can I get an allowance for this?
Pa: Help own people still want talk about money!!?? You don’t want come then don’t come la!
Me: I never said that…
Pa: Ask yall come only make so much noise, ask so much how to learn? This is good learning opportunity for you also. Just come, you cannot say no.
Needless to say, that dinner was not very pleasant. Anyway, that’s how I became a ka kia for a week in April 2025.
The week
I was brought in a few days before the whole team left for their holiday, to learn as much as I could from them. It didn’t help that my dad’s style was “learn everything yourself and don’t ask me”, while also, at the same time, “why are you so slow? Why is this one wrong?”
There was no SOP to refer to and no clear catalogue to search. The archives were a mess. I struggled to adapt to how the business ran. Emails and WhatsApp chats were printed out for my reference. Purchase orders were printed out for approval, signed, then scanned back into the system.
When the sales team left, it got worse, because I had to handle their open enquiries too. Sorting out the (literal) stacks of paper on my desk was a pure mess, and I kept losing track of tasks.
This got me thinking: how could I change any of this? My dad has always said it’s hard to keep new staff, and now I saw why for myself. Steep learning curve, little to no progression, horrible pay. In this economy? No Singaporean fresh grad would take a job like that. And because of government restrictions, without local employees it was very hard to hire foreigners, who might be more willing to work for less.
Then it struck me: why do we still need to hire in this day and age? Our workflows are fixed and repeatable. (Not perfectly consistent, since each one depends on the clients and suppliers involved, but close enough.) And fresh out of my stint in the Air Force, I remembered something my supervisor there taught me: sometimes, all you need is a vision.
And a vision I had: using AI not only to speed up work for employees, but also to replace them.
Exhibit B: a(nother) real conversation with my dad, at the end of April, after the week
Pa: So how do you feel after the one week? Not so bad right?
Me: Yea, it was okay. (No, I couldn’t say it was terrible trying to keep up. But anyway…)
Pa: They’re [the senior staff in the team] leaving soon. Like I said [in an earlier conversation, when I suggested giving them more responsibility], they’re not as invested. Not their company to fight for…
Me: Then why not use AI?
Pa: Anyone can say to use, how do you think we should use AI?
Me: Use it to help search for enquiries, clean up the internal file drive, plan delivery routes, as “admin staff”…
Pa listens.
Me: …mainly the enquiry side, then the tendering side, since you also don’t have anyone on tenders.
Pa thinks.
Pa: Mmm, just try ah. If we don’t try it’s going to be the same anyway. You can learn also.
Part 3: Ideation
I started keeping a log of my ideas and work in a Telegram channel, because that’s very convenient for me. Over the following days, I left comments under each idea: either the next step of that idea, or a rebuttal I’d thought of.
I asked around and came across platforms like n8n that I could start from. All of this while juggling my internship, scholarship interviews, fencing training, jazz lessons (hehe, this was fun) and possibly my most sociable era to date! (Meeting people before I left for the UK. We start early!)
So many different experiences inspired me too. There was a UIPath workshop I was sent for during my HTX internship, which got me thinking about agentic AI workflows. There was a YouTube video on how JP Morgan built an AI agent for investment research with LangGraph. And so much more.
Part 4: Connecting
My sponsors connected me with the right people in organisations I could get grants from. It became a flurry of pitches to different stakeholders: directors in charge of grants, engineers I wanted to learn from, and many others. I’m still applying for grants. But the best part so far was getting coffee with one of the people in charge and just asking him for advice.
Around this time, I started seeing the business in a different light. It’s literally my dad’s firstborn. He has fought hard for it for the past three decades, and he will keep fighting to keep his child alive. Suddenly, I felt more purpose. I was trying to buy time for my oldest sibling. One I had never really counted as part of the family until now.
Part 5: Inspirations, and an unlikely breakthrough
The first came from a Jane Street social dinner. How does a quant firm feeding free food to uni kids give me a eureka moment? I got talking to an engineer, and she shared how their systems were built: a platform of custom-made tools that traders could pick up as they wished. Like Lego bricks. They could update tools, toss outdated ones and add new ones whenever they liked. I was really wowed, probably because I could see myself using a tool like that. My vision shifted to fit this idea.
Then, in April 2026, I attended Jane Street’s spring week. I got to spend more time with their engineers and traders, and see first-hand how they use their system. Talking to the engineers, I realised that for a Lego-brick platform to work, the users need some agency. That was not something I could count on. I probably needed something more “idiot-proof”: something that doesn’t add to what my dad’s employees already have to juggle and learn. And frankly, their work doesn’t change nearly as much as a trader’s does, so a more straightforward, rigid system works better here.
The second came at a career fair in Michaelmas term. I was walking around, trying to find a job (yea, the market’s really rough), when I came across an exhibiting company. Let’s call them X. X was showing a product similar to n8n, and described a solution to a problem very similar to mine: workflow automation using LLMs, for large-scale financial firms. Of course, financial firms have far more restrictions than a small hardware business like my dad’s, but I saw a possible collaboration.
And I literally went:
Me: Do you guys accept client referrals?
Exhibitor A: Yea, we can give you a referral bonus, that can be done.
Me: What if the client is me?
Exhibitors A and B, both stunned.
Exhibitor B: Yea, tell us more.
So I pitched my problem statement to them. (I had been planning to build the system myself, but the Cambridge workload made it close to impossible to work on it full time. The company is basically in the ICU, and I need to buy it time, ASAP.) They kept my email, said it was an interesting problem statement, and I carried on with my job-hunting walk.
I heard nothing, and forgot all about it. Then, a few months later, they got back to me. We met up for coffee and had a very productive chat, where they explained how they had framed a new solution for my problem statement. We agreed to stay in touch. For the next half a year, I saw their product iterations, we had fruitful discussions, and I was involved in all of it.
Did I ever expect an opportunity like that from a random career fair? Not exactly. It was quite funny to me: I went there trying to find a job, and ended up giving some people jobs with my problem statement instead. Fun!
Now?
When I was back in Singapore for the summer, my dad roped in a few other vendors for us to consider, and we signed a contract with one of them to start work! (Finally!)
We’re starting with the knowledge base, so new employees have an “all-powerful” AI to ask, rather than checking in with my dad all the time. Hopefully, as we deepen this partnership, we can move our enquiry and tendering workflows entirely onto AI, and slowly become a digitalised business. (And save some trees!)
I’m glad things are moving, and I hope they keep moving on their own as I carry on with my second year at Cambridge. It’s great that I got to bond with my dad (literally a certified workaholic) thanks to my oldest sibling. And maybe, in time to come, it becomes my child too, when I inherit it.