AI Era Demands Technical Management Literacy | Generated by AI
Question: What does Tang Jie, the founder of Zhipu (GLM), say about management? How can someone manage a company without deeply understanding the technology?
Answer:
First, a small clarification: Tang Jie is the co-founder and chief scientist, not the CEO. The CEO of Z.ai is Zhang Peng. Tang Jie is primarily the technical leader and researcher. (Wikipedia)
From his recent talks and interviews, Tang Jie’s view is actually the opposite of the traditional MBA management philosophy.
He argues that in the AI era:
Management is still useful, but if you don’t understand technology deeply, you don’t even know what you’re managing. (Sohu Mobile)
Why does he think this?
His reasoning is roughly:
- AI technology changes every few days.
- New ideas (Agentic RL, inference optimization, kernel fusion, self-training, etc.) spread globally overnight.
- Falling behind is no longer gradual—it can happen in a week.
- Therefore, technology becomes the core competitive advantage.
He says that many traditional management layers existed mainly because of information asymmetry:
CEO
↓
Managers
↓
Team Leads
↓
Engineers
Managers coordinated information, assigned tasks, and collected progress.
But AI automates much of this:
- writing code
- summarizing documents
- generating plans
- debugging
- searching knowledge
- coordinating work
As AI removes communication overhead, organizations naturally become flatter. (Sohu Mobile)
Can a non-technical person still manage?
Tang Jie doesn’t say it’s impossible.
Instead, he suggests that pure management without technical understanding is becoming much less effective.
For example:
Imagine engineers discussing:
- Flash Attention
- MoE routing
- FP8 kernels
- KV cache
- speculative decoding
A manager who doesn’t understand any of these cannot judge:
- whether an idea is valuable
- why something is blocked
- whether the team is making good tradeoffs
- how difficult a task really is
So decisions become guesses.
This is different from older companies
Twenty years ago, companies like:
- Procter & Gamble
- Coca-Cola
- General Electric
could be led by executives who weren’t domain experts in every technical detail.
In frontier AI companies such as:
- Z.ai
- OpenAI
- Anthropic
- DeepSeek
the product is the technology. If leadership doesn’t understand the technology, it becomes much harder to set direction.
How do successful founders handle this?
Many successful AI founders are highly technical:
- Tang Jie
- Liang Wenfeng
- Dario Amodei
- Ilya Sutskever
They don’t stop managing people—they combine technical judgment with leadership.
As companies grow, they usually bring in experienced operators (COOs, CFOs, product leaders), while the founder continues setting the technical vision.
If you’re not technical enough
Tang Jie’s philosophy implies you should continuously improve your technical understanding rather than rely solely on management skills.
You don’t necessarily need to be the best engineer, but you should be able to:
- understand the architecture,
- evaluate technical tradeoffs,
- ask good questions,
- recognize promising new directions,
- and make informed strategic decisions.
In frontier AI, technical literacy increasingly becomes part of leadership rather than something that can be delegated entirely. (Sohu Mobile)