A small studio in Chengdu that has to live with what it ships

We build AI integrations for other people and we run two products of our own. The second half is what keeps the first half honest.

The company

Legal name
Chengdu Black Orange Information Technology Co., Ltd. (成都黑橘信息技术有限公司)
Based in
Chengdu, Sichuan, China
What we do
AI workflow integration for small teams, plus app teardown reports through AppXray.
Languages
English and Chinese, including the handover documents.
Contact
blackorange0425@gmail.com, or the contact form. We reply within one business day.

How we build and run things

These are not aspirations. They are the rules our own automation runs under, and you can ask us to show you any of them.

Plans live in files
Work is written down before it starts: what will be done, what done means, and which paths it is allowed to touch. A plan that only exists in a conversation is not a plan.
Every write takes a lease
Before touching a file, work claims it. If another worker holds an overlapping claim, the second one is refused. Two writers racing on the same file is a bug we designed out, not one we clean up after.
Every action is in the ledger
Our operations append to a hash chained audit log: what ran, when, on whose instruction, and what it changed. Nothing irreversible happens without a line in it.
Spending has a daily ceiling
Unattended automation stops at a fixed budget for the day. When it is hit, work stops rather than continuing quietly.
Reuse before building
If an existing tool covers most of a need, we use it and note what we changed, where it came from and what it costs to maintain. Rebuilding something that already exists is a choice, and it gets argued for like one.
Declining is allowed
If a job needs something we have said we do not do, the answer is no, early. A project we should not have taken costs you more than a project we turned down.

Why the name

Black is the engineering half: stable, quiet, focused on the parts nobody sees. Orange is the product half: visible, energetic, willing to try the thing that might not work. A company that is only the first half ships nothing. Only the second half and it ships things that fall over.

Want to see the standards applied to your workflow?

Start with the diagnostic. It is the cheap way to find out if this is worth doing.