EfolusiWe build software, then run it ourselves.
Efolusi is a small, self-funded software company from Indonesia. We started it to build the tools we kept wishing we had, and we've kept going the same way ever since: find the problem that annoys us, build the product, keep it running. The legal name is PT Efolusi Dunia Teknologi, in case you need it for partnership paperwork.
We take on a lot, on purpose.
We're not a one-product company. When a problem keeps getting in the way, we build a product for it and use it ourselves.
The rule is simple: everything has to be good enough that we'd use it ourselves. If it isn't, it doesn't go out.
Built in Indonesia, used everywhere. Open our products anywhere and they work the same.
Founder-led and hands-on.
Decisions stay close to the products. Both are hands-on, making sure every product runs the way it should.
Proof beats promises.
One principle runs through everything here: we only build products that are useful to us and likely useful to someone else. Each one is told the same way: the problem, what we built, and what you can check right now.
Most AI assistants wait to be told every single step, which turns them into a second job instead of actual help.
An agent workspace that takes a goal, works out the steps itself, and learns your context as it goes.
Active beta at zoyya.xyz. Zoyya acts, you watch: every run is visible and reviewable.
Running a small company means the same operational work every week, and nobody left to hand it to.
A control plane where teams build, run, and govern AI agents across the org, the platform layer products like Zoyya build on.
In build, run on our own operations first; external availability is not claimed yet.
Every product needed its own keys, quotas, and billing for every LLM provider, and none of it was comparable.
One control plane in front of every AI provider, with a single key, one set of limits, and one bill.
Live as the gateway all Efolusi products call for AI; provider routing and spend are visible in one place.
One long video actually holds a week of content, but cutting it by hand still eats up an entire evening.
A content clipper: long video in, short captioned clips out, with a review pass before anything ships.
Being validated as a paid service for agencies and content teams; not represented as a mature product.
Small file jobs like converting and compressing always end up on some ad-filled, tracker-heavy site you have never heard of.
A collection of file utilities that run directly in the browser without an account.
Live at toolips.xyz with a free tier and a paid Pro plan.
Running a grid by hand means orders go in late: when the price swings up or down hard, the profit opportunity is already gone.
Automated grid and DCA trading bots with manual or model-assisted configuration.
In revenue validation with legal, credential-security, and auditable-performance gates. The only trading product we make.
The same content had to be pasted into platform after platform, every single day, and something always got missed.
A social network where you write once and cross-post to every social account you connect.
Being repositioned and tested for revenue; product-market fit is not yet proven.
Work started before anybody agreed what was being built, so the rework arrived later and cost more.
A specification workspace where PRDs and specs are reviewed and versioned before development starts.
In build and used on our own products first: the spec comes first.
Giving an AI assistant useful context usually means handing it your secrets in plain text.
A vault and private knowledge system your AI can use but never read.
In build; the encryption model is the product and is being reviewed before any public claim.
Our own infrastructure was spread across servers and dashboards, and every incident started with a search.
One dashboard for every server, database, URL, and certificate you own, with monitoring attached.
Being built as a shared operations layer; cross-product adoption is not claimed before it is evidenced.
Every product we started needed the same buttons, forms, and colors, and a small team cannot afford to redesign them a dozen times.
A design system for reusable components, tokens, and interface patterns, published as an npm package.
Published as @efolusi/meridian under the MIT license; its contribution to product consistency can be inspected directly.
The company on paper.
See what we've built.
The same care goes into everything we ship. Go find the one that fits your work.