Kai
Cheng

Somewhere around his 200th enterprise AI deployment, Kai Cheng noticed a pattern. The client would get a shiny new AI tool. The AI would wait. The human would come back, open the tool, type the prompt, click run, copy the output into a spreadsheet, and close the tab. Repeat. He'd built some of the most sophisticated production AI systems in the industry - and people were using them like slightly smarter search bars.

That is the specific frustration that launched Creao AI. Not a gap in model capability. Not a missing dataset. A human still sitting in the loop, serving as the CPU for a machine that was supposed to replace the CPU. Cheng saw it at a company designing supply chain logistics. He saw it at a Fortune 500 firm automating customer service. He saw it everywhere. The bottleneck was never the AI. It was the hand-off.

Creao AI launched commercially in September 2025. By April 2026, it had 200,000 users - none of them acquired through paid advertising. Every single one arrived because someone showed it to someone else. The product spread the way good tools used to spread, before marketing departments existed: by working.

The platform operates on three layers. First, a coding agent that turns plain-language descriptions into fully functional tools - no programming required. Second, Agent Apps: those tools, running on schedules, responding to triggers, executing workflows without waiting for a human to hit play. Third, a central workspace where humans oversee the whole operation, steer the system, and watch the work happen. Not a chatbot. Not a co-pilot. A factory floor with AI on every station, and the human as plant manager.

Cheng's own team of 22 runs influencer marketing, SEO pipelines, and content production entirely on the CREAO platform. They are, by any reasonable metric, operating with the output of a team five times their size. The company doesn't just claim this - it uses itself as the benchmark. Every internal failure is, in Cheng's words, a product insight. Every broken workflow is a feature spec.

The funding story is almost as compressed as the growth. A seed round, then a Pre-A, then a $10 million round led by Prosperity7 Ventures - the investment arm of Aramco's $3 billion fund - all within twelve months of founding. Total raise: $25 million. Earlier investors included GL Ventures, HongShan, Yunqi, Hua Capital, and Grishin Robotics. The pattern of the cap table mirrors the pattern of the product: fast, global, no filler.

That philosophy - execution as the only real moat - shapes everything Creao does. Cheng is not betting on a proprietary model. He is not betting on a first-mover advantage in a category that didn't exist two years ago. He is betting on the compounding effect of a team that has to run its operations on its own product, in real time, with real stakes. The best AI companies, he argues, won't be the ones with the most elegant pitch. They'll be the ones who figured it out in production, under pressure, with their own money on the line.

Before any of this, Cheng spent a decade in the unglamorous trench of enterprise AI implementation. Not research. Not demos. Production systems. The kind that have to work on a Tuesday at 3 a.m. when no one is watching. More than 250 clients. He studied statistics at Columbia - MA, 2013 to 2015 - and spent the years after that learning what statistics looks like when it collides with real organizational behavior, real messy data, and real humans who don't read documentation.

The Creao thesis is, in a way, the thesis he has been developing across those 250 engagements: AI is not a feature to be bolted onto existing software. It is a new operating layer. Agents are the primary users. Humans are the architects. "Adding AI to old software is a dead end," he wrote on X in late 2025. "We're not patching the old system. We're building the foundation for this new world." The statement would sound grandiose if the traction didn't make it sound, instead, like a progress report.