AI Agents & Automation
MANUS: AI Agent ล่าสุดจากจีน กำลังท้าทายทุกความเชื่อเกี่ยวกับ AI!
Summary
What separates Manus from ChatGPT and DeepSeek is that it does not answer — it completes the task.
The talk demonstrates it live: planning a trip, writing software, building a game.
The consequence at issue is what agents do to the future of work, which is a different question from what chatbots do.
Key points
Answering and doing are different tiers
A tool that finishes the task redefines the work rather than merely speeding up the existing version of it.
That it is Chinese is part of the story
Silicon Valley’s concern is not only about capability but about who built it.
The demonstration is the strongest evidence
Claims about AI capability are hard to check, so showing the thing running outweighs a benchmark.
Accountability follows immediately
Once the system acts on its own, who answers for the outcome stops being a theoretical question.