*Every meeting tool assumes you speak one language.
*
Otter.ai. Fireflies. Notion AI. Microsoft Copilot. They are all built for a boardroom in San Francisco where everyone speaks English, with a slight accent at most.
That is not how Southeast Asia works.
I grew up in Indonesia. Every meeting I have ever been in has been a mix. Bahasa Indonesia when we are relaxed and talking fast. English when we are pitching or presenting. Mandarin when someone senior from a Chinese-owned conglomerate joins the call. Sometimes all three in the same sentence.
This is not unusual. This is just how ***business* gets done here**.
When I started looking for a tool that could handle this, I found nothing. The best I could do was run Otter.ai and watch it struggle with every Bahasa sentence, every code-switched phrase, every moment where the conversation naturally moved between languages. The transcripts were unusable. The summaries were worse.
That was the moment I knew Bysik had to exist.
The gap nobody was talking about
Southeast Asia has 670 million people. Indonesia alone has 270 million. The region's digital economy is on track to hit $1 trillion by 2030. Startups are scaling fast, enterprises are expanding across borders, and remote and hybrid work is now the default for any professional team.
And yet every single AI meeting tool was built by American companies, for American users, with English as the only language that matters.
The few tools that claimed multilingual support were doing post-processing translation. You would record your meeting in one language, wait for it to be processed, and get a translated transcript hours later. That is not translation. That is transcription with a dictionary attached.
Real multilingual work is real-time. It switches mid-sentence. It does not wait.
_
Why on-device changes everything
_
When we started building Bysik, we made one architectural decision that changed everything else: the AI model runs on-device.
No external API. No audio routed through a server in Virginia. No third-party dependency.
This was not just a privacy decision, though it matters enormously for Indonesian companies navigating data localisation requirements under PP 71/2019. It was a product decision.
Running the model locally means Bysik works offline. It works in a Jakarta office building with inconsistent WiFi. It works on a laptop on a flight between Taipei and Singapore. It works in any environment where cloud-dependent tools fail.
It also means our unit economics are fundamentally different from every competitor. No per-minute API cost. No usage-based billing that punishes power users. A flat, predictable product that scales without our costs scaling with it.
What we are building
*Bysik is an AI meeting intelligence platform with four core features: real-time multilingual translation, automatic meeting notes, speech-to-text, and a productivity suite we call Launch Pad.
*
The real-time translation is the feature I am most proud of. You open Bysik before a meeting. As people speak, regardless of whether they switch between Bahasa, English, or Mandarin, subtitles appear in real time. No delay. No post-processing. Just the meeting, understood.
We are post-beta. We are approaching launch. Indonesia is our first and primary market.
Why now
AI adoption in Southeast Asia is accelerating faster than most people outside the region realise. Indonesia's AI startup funding grew over 200% in the first half of 2025. The enterprise market is hungry for tools that actually fit how they work, not tools built for someone else that they have been told to adapt to.
We are not trying to be a cheaper Otter.ai. We are building the meeting intelligence tool that Southeast Asia deserves. One that was built here, for here, from the ground up.
If that resonates with you, follow along. We are just getting started.
*Bysik is currently in pre-launch. Visit bysik.app to learn more.
*
Top comments (0)