Pragmatik draws backing for digital and physical AI agents
What's the deal? PragmatikDealroom has a profile for this one. Try Dealroom →, founded by Lin Junyang — the former head of Alibaba's Qwen large language models — has raised an early-stage round led by Sequoia China and Gaorong Capital. Tencent and the Shanghai Future Industry Fund also participated.
What's the endgame? The company is building a new generation of AI agents that span both digital and physical worlds. Its site describes "digital agents" for knowledge work and enterprise operations, and "physical agents" that act inside robots and real-world devices.
Pragmatik frames its ambition around taking research to product and enabling "long horizons" — agents that pursue a single goal over minutes, hours, or longer. Its hiring page names the goal outright: superintelligence.
Why now? The launch comes five months after Lin's exit from Alibaba. He joined Alibaba's DAMO Academy in 2019 and later led the Qwen series from zero, building its models and open-source ecosystem into one of the world's most visible open LLM families.
What could go wrong? Lin's track record is with foundation models, not deployed agents. Whether he can build something at Qwen's scale outside Alibaba remains untested, and the general-agent field is already crowded.
The signal: In today's AI market, talent has become a strategic asset in its own right. Models can be retrained, compute bought, and data accumulated — but few people have shepherded a top-tier foundation model through pretraining, post-training, open-sourcing, and large-scale engineering. Backing Lin's new venture is a bet on that scarcity.
Read more: 36Kr
Image credit: Generated with Gemini
