There is a tendency to think about business innovation as focused on technological change and the new business models to bring that change to customers in order to serve them better. That’s certainly the entrepreneurial mindset of most companies and is typified by the tech giants of Silicon Valley.

In the AI community, which is a highly active field of innovation at the moment, the conversation is often phrased in the terminology of the frontier of change, and frontier models, referring to the stair-step breakthroughs to new levels of performance which seem to occur with rapid frequency.

When the Chinese AI firm called Deepseek recently unveiled its latest model, it was evaluated in that context: was it a new technological frontier? The general opinion is no – it represents some surprising advances in the quality of output achieved relative to the cost of development and what the AI community calls “training”, but hasn’t reached a new performance frontier.

Here’s what a lot of analysts seem to be missing, however. Deepseek may represent breakthrough innovation on a different and possibly more important frontier, that of organizational design. The Deepseek model was developed by a business unit with

no management,

no business administration,

no resource allocation methods,

no levels and no hierarchy,

no structure,

no fixed roles,

no titles,

no KPI’s,

no targets, and

no politicized competition to climb the corporate ladder.

All of the traditional trappings of Western business administration have been thrown out.

This suggests that there is a new frontier of innovation, that of organizational design and business management.

Deepseek’s replacements for administration are manifold and represent a completely new system.

Freedom

In place of business administration, Deepseek’s founder speaks of freedom – freedom to pursue exceptional accomplishment. This is the enabling incentive for individuals, teams and the firm. No one can plan in advance where this freedom will lead. There is only the beacon of exceptional accomplishment to aim for, without any false quantification of what that might be. Will people be happy with what they accomplish? Freedom enables them to find out.

Spontaneous division of labor

How is freedom organized? It isn’t. At Deepseek, there is a spontaneous division of labor in which individuals create ideas, teams form and unform in developing those ideas, and cross-team collaboration occurs as it’s prompted by the evolution of ideas. Individuals are recruited to teams, or they migrate to a team that they’re aligned with. There’s no HR Department, no job description, no bosses with instructions on how and where and when to work and with whom.

Discovery versus targets

The traditional Western form of management includes the setting of targets against which to measure outcomes. Individuals are given targets, teams are given targets, firms set annual targets. Deepseek eschews this form of management. The alternative to plans and targets is mission-driven discovery. The mission is AGI, the pathway to it is uncertain and unmapped, and the method of progress is idea-inspired experimentation to discover possible next steps. Discovery is action-driven, not target-driven.

Teams form around ideas

Teams have become a central tenet of advanced organization, built on a project management mindset. Teams are assembled around project goals, with the expertise to address identified project needs and the goal of completing the project. Deepseek assembles teams around ideas. Any individual can originate an idea and the team assembles around exploration and experimentation to discover where the idea might lead. It might lead nowhere, in which case the team can reassemble around another idea, or it might lead somewhere unexpected, in which case the team can reconfigure. Teams can collaborate across boundaries, so a team is not an exclusive concept but a self-assembling stream of expertise that can flow in multiple directions.

Transparency and knowledge-sharing

Conventional management concepts assume that specialized knowledge can and should be held and focused in departments, areas of expertise, or dedicated projects. At Deepseek, the preference is for total transparency and unfettered knowledge sharing. Every individual has access to all the firm’s knowledge without restriction. Know-how is accumulated and shared as a firm-wide asset.

Resources versus resourcefulness

One of the traditional tasks of management is resource allocation – deciding who gets to utilize the firm’s resources. At Deepseek, every individual and team has access to the firm’s resources, without permission or planning required. An idea can command the GPU’s and compute time needed to validate it. The productivity variable then shifts to resourcefulness versus resources.

It appears that Deepseek was able to develop their AI models effectively despite restrictions on their access to Nvidia chips – which is evidence of their system’s resourcefulness in the context of limited resources. Their Mixture-of-Experts architecture, while not new, has been impactful in cost reduction and efficiency, leading to a resurgence of interest in this kind of model.

Eliminate the politics of competition

To “climb the corporate ladder” in a Western corporation is to engage in a political competition with peers to gain recognition, assert superiority and to individually break away from the pack. The Deepseek model – and perhaps the Chinese model – is to aim for collective effort and collective impact. The collective wins via exceptional achievement in a context of mission alignment and shared purpose. Teams share insights openly in order to accelerate innovation for the group as a whole. The kind of political competition that can destroy cultural coherence is not admitted.

Passion over Experience

Deepseek founder Liang Wenfeng has indicated a preference for youthful passion over experience, whereas the latter is usually key to hiring preferences in traditional management. Experience, he says, brings baggage that can weigh heavily on open-ended exploration. For short-term goals, hiring experienced individuals makes sense. But long-term success does not depend on past experiences. Rather, it depends more on foundational skills, creativity and passion.

The New Frontier.

In Liang Wenfeng’s view, organization is not just an innovation frontier, it’s the ultimate source of competitive advantage. He said, “In disruptive tech, closed-source moats are fleeting. Our real moat lies in our team’s growth – accumulating know-how, fostering an innovative culture”.

KFSOs

The Deepseek organizational model may point in a direction Western firms should follow. It represents a different management culture, disruptive organizational thinking and an iconoclastic approach to conventional wisdom. The Kinetic Flow State Organization design that Mark Beliczky and I have encouraged firms to pursue is directionally aligned with Deepseek. Deepseek’s willingness to freely explore in multiple directions and adjust rapidly to feedback generated by continuous experimentation is highly kinetic. Its knowledge flow is accelerated by the removal of barriers, including management layers, planning, and resource allocation protocols. Its passionate young researchers achieve personal flow states in their energetic commitment to the challenging task of searching for a pathway to AGI.