From Unlimited Tokens to Full-Agent: MiniMax's AI Native Organizational Evolution
MiniMax, an AI startup focusing on multimodal models, went public on the Hong Kong Stock Exchange in January 2026. The company adheres to a dual strategy of large models + applications and ToC + ToB. Internally, it provides unlimited tokens to all employees, uses agents to automate workflows, and targets high-value tasks that humans dislike, significantly improving efficiency and flattening the organization. In the next 2-3 years, AI will deeply integrate with various industries.
At the 2026 China AIGC Industry Summit, Hu Weiqi, head of MiniMax's ToB commercialization in China, shared the company's practices and insights on building an AI-native organization. Founded in 2023, MiniMax is a startup focused on multimodal models—text, video, voice, and music—and went public on the Hong Kong Stock Exchange in January 2026. From its inception, the company envisioned the next generation of AI: an intelligence that passes the Turing Test, delivers extreme customer experiences, and addresses the limitations of small models, such as poor generalization and customization issues. This vision is encapsulated in their mission: 'Intelligence with Everyone'—making AI affordable and accessible to all.
To achieve this, MiniMax has consistently pursued a dual strategy of advancing both large models and applications, while balancing ToC and ToB efforts. Their product portfolio includes the text model MiniMax-2.5/2.7, the Hailuo video generation model, voice models with human-like realism, and a music model. These models underpin both consumer products like Xingye (emotional companion) and Hailuo AI, as well as enterprise offerings such as the MaxClaw agent platform. Notably, the company emphasizes dogfooding—using its own models internally to refine them.
Internally, MiniMax has pioneered an AI-native organizational culture. Since September 2024, all 400-plus employees have been given unlimited token quotas to use tools like Cursor, and by August 2025, agent projects were implemented across functions. For instance, HR agents handle resume screening and interview scheduling; developers use AI for coding, with human review only at commit; and marketing agents generate promotional videos and track sales leads. The key insight: start with high-value tasks that humans are least willing to do, which minimizes internal resistance and boosts satisfaction. This approach has led to a 30% R&D efficiency gain, with 30-50% of code auto-generated.
Hu Weiqi noted that AI flattens organizations. The boundary between front-end and back-end R&D blurs; product managers now create demos directly, and developers focus on production-scale implementation. The consumption of tokens has become a new metric for efficiency, reflecting individual and team AI adoption levels. Looking ahead, Hu believes that over the next 2-3 years, AI will rapidly integrate with every industry, reshaping productivity tools, business models, and organizational structures. 'The curtain on AI has just risen,' she said, encouraging enterprises to stop worrying and start experimenting with AI directly.