We Didn’t Start with AI. We Started with Frustration

When we first started ARSA Technology, we weren’t thinking about neural networks or computer vision. We were thinking about long lines at clinics. About coal trucks that overloaded because no one was really watching. About factories where “monitoring” just meant someone squinting at a CCTV screen all day.

We started with frustrations — the kind that everyone notices but no one feels responsible to fix.

So we made ourselves responsible.

Building in the Margins

In Indonesia, technology often arrives late. By the time it gets here, it’s usually imported, overpriced, or irrelevant. So instead of waiting for Silicon Valley to notice us, we decided to build here, for here.

At first, that meant scrappy demos with hacked-together sensors. PowerPoint decks that oversold our capabilities. Long nights training AI models on borrowed GPUs.

And clients who said, “Kameranya bagus, tapi bisa nggak ini nyala di site yang nggak ada sinyal?”

Valid point.

We didn’t just need smarter tech. We needed sturdier, simpler, more contextual solutions. That’s how ARSA’s DNA formed: part R&D lab, part field team, part infrastructure realist.

The Hardest Problems Are Human

You’d think the hardest part of building Vision AI is accuracy. Or edge deployment. Or dataset labeling.

It’s not.

The hardest part is getting people to trust a camera that thinks.

Especially when the camera decides whether a truck gets to enter, or if someone is wearing proper PPE, or if a self-service kiosk thinks your face looks “healthy enough.”

So we learned to design not just for intelligence, but with empathy. Our dashboard doesn’t just stream alerts — it tells stories. Our kiosks don’t just scan — they greet. Our algorithms explain themselves, when needed.

Because trust is earned, especially when you’re asking humans to work with machines.

We’re Not Here to Disrupt. We’re Here to Work.

There’s a fetish in tech for “disruption.” But most of our clients don’t want disruption.

They want the conveyor belt to stop jamming.
They want the data to be real.
They want fewer accidents on shift.
They want a partner, not just a vendor.

At ARSA, we’re proud of the fact that many of our pilots turned into long-term deployments — not because we were perfect, but because we were present. We picked up the call when systems went down. We adjusted models when the lighting changed. We showed up.

That’s what building deep tech in Indonesia looks like: less hype, more heat-resistant enclosures and WhatsApp support.

What We’ve Learned (So Far)

• AI is not a product. It’s a commitment.
• Real innovation isn’t glamorous. It’s relentless iteration.
• Every successful deployment begins with a human conversation, not a pitch deck.

We still have a long way to go. But we’ve learned to celebrate progress — every alert that prevents a collision, every kiosk that eases a nurse’s workload, every plant manager who says, “ini beneran bantu kerjaan saya.”

That’s why we build.

Not to chase trends, but to solve what matters. Locally. Sustainably. Respectfully.

This article is part of Machine State — ARSA Technology’s official publication exploring intelligent systems and future tech.

Written by Hilmy Izzulhaq
Founder @ ARSA Technology — 7 years building AI Vision & IoT solutions in heavy industry, parking, and smart city.