NeuxMind.AI, the Chinese enterprise AI-agent company behind the Shenshou agent operating system, closed a $27.5 million Series B round led by Saize Capital on November 17, 2026. Existing dollar- and renminbi-denominated shareholders also participated. The company did not disclose a valuation, but said annual revenue roughly doubled in 2026 to approximately $27.5 million, with overseas revenue topping $13.8 million.
Founded in 2020 and led by Zhang Lei, NeuxMind.AI sells "Agent Workers" to banks, financial institutions and government bodies across the Middle East, Southeast Asia, Japan and Greater China, under a results-based pricing model tied to revenue gains, cost cuts or efficiency improvements rather than traditional software licenses.
From digital agents to physical robots
What sets this round apart from a typical enterprise-software raise is where NeuxMind.AI is pointing the money next. The company says it is acquiring a robotics team and plans to extend its Shenshou agent brain beyond office software into physical hardware, building dispatch and intent-recognition systems for drones, robot dogs and robotic arms, with products targeted for the first half of 2027.
That would turn NeuxMind.AI from a back-office AI vendor into a company supplying the decision-making layer for embodied machines, a strategy that mirrors how several enterprise-AI players are chasing physical AI as a new growth frontier rather than building robots themselves, according to Dealroom.
Traction and business model
NeuxMind.AI says it has recorded zero failed outcome-based deliveries so far, a claim that matters because its revenue is contractually tied to whether its agents actually hit the business outcomes clients pay for. Founder Zhang Lei has warned that "charging for fuzzy outcomes is digging yourself a hole," suggesting the company is deliberately avoiding vague success metrics as it scales into new verticals and now into hardware.
Investors and competitive landscape
Saize Capital led the round alongside unnamed existing backers, continuing support through the company's growth from a domestic enterprise-software vendor into an international agent platform. NeuxMind.AI's move into physical AI puts it in a different lane from most enterprise-agent rivals, which stay confined to digital software, and closer to the broader embodied-AI race where labs are racing to build the control systems that let robots take instructions from language models.
What to watch next
The real test will be whether NeuxMind.AI can ship a working robotics dispatch product by its stated H1 2027 target, and whether banks and government clients that trust it to run back-office agents are willing to extend that trust to physical hardware like drones and robot arms operating in the field.