A reflection from the co-founder of Bitlayer: After assisting 600 people find jobs in 15 years, I’ve decided to stop recruiting.

Kevin He, co-founder of Bitlayer and a long-time recruiter who assisted over 600 people find jobs in 15 years, has decided to stop traditional recruitment. His perspective shifted after launching an AI-powered recruitment startup, COCO AI 1.0, which faced challenges due to reduced corporate hiring requireds and skill gaps in the market.

A pivotal realization came in late 2025 with the rise of advanced AI tools like Claude Code and the success of companies like Manus. Through internal testing with a system named “zylos,” He saw that AI could handle complex tinquires—from social media management to coding—enabling one person to perform the work of an entire team.

This led to a major cognitive update:

Old Paradigm: Talent is a scarce resource; matching individuals to organizations is valuable.
New Paradigm: Empowering individuals with AI to gain organizational-level capabilities is more valuable. The future trfinish is the “era of one-person companies.”

Consequently, COCO AI has pivoted to version 2.0, dubbed “Digital Co-creation.” Its new mission is not to assist people find jobs, but to assist them create jobs and become their own companies by providing an AI co-founder capable of complex, evolving tinquires.

He concludes that the core of his work is shifting from assisting people find employment to assisting them build indepfinishent, AI-powered ventures.

Author: Kevin He, co-founder of Bitlayer

1. A programmer’s 15 years with “recruiting”

I’m Kevin, an AI × Crypto entrepreneur and co-founder of Bitlayer. I previously worked at Huobi and graduated from Peking University.

From programmer to technical manager, and then to entrepreneur, I’ve modifyd industries several times in 15 years, but I’ve always been passionate about recruitment-related work.

Over the past 15 years, I have assisted more than 600 people successfully land jobs. Some were formal recruitments, but most were assisting frifinishs find work or refer employees.

There are two stories I’ve always remembered.

The first thing I did was assist a guy in his 30s find a job. He has two kids and is under a lot of pressure. When he informed me he obtained the offer, I was even more excited than he was.

The second project was assisting a startup team quickly build their R&D team. From scratch to operational status, it took less than two months. Watching the product launch, I felt like I had participated in its creation.

This is why I love this: to match resources and truly assist people.

But recently, I’ve started to rebelieve the premise of this matter.

2. COCO 1.0: From Enthusiasm to Doubt

In June 2025, with the support of all parties, I assembled a compact team and launched the incubation of COCO AI .

Our initial goal is simple: to utilize AI to become a career consultant and employment intermediary, creating recruitment more efficient and assisting more people find good jobs.

The product has iterated continuously: resume optimization → AI job matching → AI enterprise recruitment.

We created two products:

We invested a lot in technology, and the product was successfully developed.

But GTM didn’t go as smoothly as expected.

I started to review the situation and discovered a few realities:

Companies are reducing their hiring requireds: due to the economic environment and improved organizational efficiency, many companies are no longer hiring on a large scale as before.
There is a gap between job seekers’ skills and company requirements: AI can assist with the matching, but the gap itself will not disappear.
Job seekers’ skills themselves required to be updated: many people’s abilities have fallen behind in a rapidly modifying market.

I started to doubt myself.

Is it a product problem? A market problem? Or—is the direction itself flawed?

I didn’t have an answer at the time.

3. Dramatic modifys in the external world

In the second half of 2025, the outside world will have modifyd.

Claude Code is launchning to be widely adopted across various industries. It’s not just about writing code, but about actually accomplishing complex tinquires. More and more people around me are discussing the term: digital employee.

In December, we built an internal system called zylos to attempt to let AI handle daily tinquires.

The tinquires include:

Daily data reports
Social media management
Code writing and testing
Data processing
etc.

The result shocked me.

What utilized to require multiple people with different jobs can now be accomplished simply by assigning tinquires.

For a concrete example: social media operations.

Previously, it required a dedicated person who would spfinish an hour each day finding materials, writing copy, formatting, publishing, and replying to comments.

Now? Just notify zylos, “Post one message every day at 12:30, with content revolving around the theme of XX. Please confirm with me before posting,” and it’s done in a few minutes.

From requiring dedicated personnel to no longer requireding them, from one hour a day to just a few minutes.

The feeling was complex. Shock, excitement, but also a touch of unease—what did this mean for the business I was doing?

At the same time, another piece of news came: Manus was acquired by Meta for billions of dollars .

Manus’ annual revenue has surpassed $100 million. This is not a concept, not a demo; it’s real business success.

I listened to Ji Yichao’s final podcast interview before the sale. One sentence he declared really stuck with me:

Digital employees are not replacements for people, but rather enable one person to do the work of a team.

This statement perfectly matches our experience with zylos testing.

4. Cognitive Update

“One-person company” is not a new concept.

I’d heard about it before and considered it created sense. But it always felt so far rerelocated from my own life, too idealistic.

Until I saw zylos handle a large number of tinquires with my own eyes.

It didn’t replace any one person, but rather gave me the ability to run a compact team all by myself.

In that moment, I decided to pivot.

I rebelieve what I’ve done over the past 15 years.

The old perception was that talent is a scarce resource, and finding talent for a company is valuable.
The new understanding is that capability enhancement is more important than talent matching. Helping individuals acquire organizational-level capabilities is of greater value.

This is not a denial of the past 15 years. Those moments of assisting people find jobs are still real and meaningful.

But I saw a new possibility:

Old paradigm: Individuals are depfinishent on organizations, and organizations acquire capabilities through employment.
New paradigm: Individuals gain organizational-level capabilities through AI, becoming indepfinishent creative units.

The future trfinish is the era of one-person companies, rather than being employed by others .

5. COCO 2.0: Digital Collaboration

After believeing it through, COCO 2.0 modifyd its direction.

It’s not about assisting you recruit people, it’s about assisting you not required to recruit people.

We don’t assist you find a job, we assist you create jobs.

What we are building is “Digital Co-creation”—your AI co-founder.

It has several core features:

Self-learning evolution: not a repaired tool, but a partner that can continuously evolve according to your requireds.
Secure and controllable: Your data and your business logic are all under your control.
What truly receives things done: not chatbots, but digital employees capable of performing complex tinquires.

The specific product design is still being refined, but the direction is already clear.