Solo AI studio

Turning ideas into things that work.

First product KOL Engine is live and taking briefs.

AI products, built solo

Turning ideas into products you can actually use.

ZIV-AI is where I build on my own. I use AI to turn my own ideas into products people can use, one at a time. The first one is live now and taking briefs.

The first is KOL Engine: send a Douyin campaign brief, it shortlists the right creators, and media ops follows up. I built it with vibe coding and automation. Notes on how it gets made live here too.

projects

Project on ZIV-AI

On ZIV-AI I turn what I build into products you can actually use. One is live now and taking briefs.

KOL Engine

Douyin creator discovery and campaign delivery: submit a brief, the engine shortlists creators, media ops delivers. Service runs in Chinese.

Next project

In the works

article

Insights

I said I knew nothing about testing. My repo had 342 of them.2026-07-07 All posts →

KOL Engine · live, taking briefs

Fast shortlists. The grunt work goes to the system.

Send your campaign brief. Backed by a maintained creator library and structured evaluation, we deliver an actionable Douyin creator shortlist — the machine filters, people decide. Service runs in Chinese.

Submit a brief

Why it is fast

The system filters, people decide

A library that compounds

No brief starts from zero. Creator profiles, works data and past collaboration records accumulate in the library and carry over across projects.

Structured evaluation

Each brief becomes explicit criteria; every candidate is assessed for fit with risks flagged — consistent and traceable. Calibrated against real client picks over two back-test rounds: creators in the recommended tier were chosen at roughly twice the rate of the rest.

People make the calls

Brief confirmation, candidate removal and final selection stay human. Media ops reviews every candidate before the list reaches you; the final picks are yours.

How it works

How a shortlist gets delivered

Brief structuring

Your brief (text or file) becomes explicit filters and watchpoints.

Candidate pool

Pre-screen from the library, plus fresh discovery on Douyin to fill the gaps.

One-by-one evaluation

Structured fit assessment per candidate, risks stated.

Review & delivery

Media ops removes the misfits; you get a shortlist with reasons per creator, tick your picks, and we follow through.

1,200+

creators in the library (growing)

3,000+

structured evaluations to date

2 rounds

of real-list back-testing

3 days

typical turnaround

* Figures shown are demo-calibre; actuals vary by project.

Your next shortlist starts with one brief.

A form, a guided chat, or just drop the brief file — all three work.

Submit a brief