The last six months we’ve felt a quantum shift at Weaviate. The buzz around generative AI has greatly increased interest in AI-native tech. More and more people realize that a vector database like ours may be nearly essential for most enterprise applications of LLMs. Meanwhile our core open-source technology has passed the MVP stage and new instances of product-market fit are occurring so fast that it’s a challenge to keep up with them.
Hence that quantum shift; now, our challenge isn’t proving that there’s a role for Weaviate in the coming AI disruption, it is determining how best to scale our operation.
As a founder, this is a good problem to have! In many ways, it is the very problem Etienne and I set out to have when we created our company in 2019. But shaping the growth of a company is very different from shaping a product. It calls for a different kind of vision because people aren’t code, tools, or processes.
Rather than envision departments and hierarchical structures, I naturally gravitate to a more organic approach–a corporate equivalent of “cell theory”. As far as I know, this biological simile was first described by a Dutch tech entrepreneur named Eckart Wintzen, who wrote about it in his 2007 memoir, Eckart’s Notes.
Wintzen was admittedly idiosyncratic but also undeniably successful. His company, BSO/Origin, grew from a one-man operation in 1973 into a 10,000-person multinational with $1B in revenues and a remarkable record of consistent profitability before he sold a majority stake to Philips in the mid-’90s. Throughout BSO’s period of rapid growth, the company was able to remain lean and entrepreneurial because it reproduced by forming new “cells”.
Why cells? And when should they divide?..
Virtually all living organisms are made up of cells. As organisms grow and develop, they follow two basic rules: The first is that there is a fundamental limit to the size of an individual cell, beyond which it cannot function efficiently. So, growth is not achieved by having individual cells get bigger and bigger; instead, cells divide and replicate. The second is that as organisms develop, cells specialize to fill individual functions.
Wintzen seems to have had the first inklings of his cell theory when his company reached 100 people. At that moment, his inspiration was to split it into two, with one 50-person cell under his control and another under the control of his second-in-command. (He came to believe that the maximum team size for optimum efficiency was in the 50-70 person range; there’s no minimum size.)
Although Weaviate has not quite reached the point where size alone dictates a cell division, we have already created new cells in order to follow that rule of cell specialization.
Each cell has a singular focus
For example, we used to have a single team devoted to “Solution engineering & customer success”. As we’ve grown to put more demands on that team, it’s become clear that those are actually two distinct (if related) assignments. So, we’ve split that into two cells: Sales Engineering, and Customer Success.
Similarly, as the number of developers using Weaviate has grown, we’ve realized that it now makes sense for DevRel to split into two specialized cells. So now we’ll have a Developer Growth team–a cell charged with increasing the number of developers using our open-source software– and a Developer Relations cell devoted to engagement with the community. They’ll ensure that our documentation is complete and develop the educational materials needed to keep our developer community happy.
It will be a while before any of our specialized teams exceed the threshold size for efficiency, for example when we have 50+ people in Developer Relations alone. But there will be more cell division prompted by specialization.
Although I can’t predict exactly how that will go, I can easily describe the algorithm we’ll use to decide when it’s time: We should split a team into two (or more?) cells whenever it holds a team meeting and a significant number of people find themselves thinking, “Why am I even listening to this? It doesn’t concern me.”
Making it work
Per Eckart’s Notes, there are a few key principles to follow to help ensure the success of this cellular approach to growth. Perhaps the first principle is that when a cell splits, both new cells should be led by someone from within the organization and if possible, from the “parent” cell. That way, cell leaders always have the essential organizational knowledge and can be guardians of the culture.
Generally, the leader/coach of each cell has freedom to structure their team as they see fit, in order to best achieve their cell’s goals. Individual cells should have a minimal (or ideally, no) administrative staff or “departments”. The cell leaders collectively form a management team, and of course they must be a team so that the company can share a unified vision and present a unified brand.
Growing Weaviate for the future
The cellular model described by Eckart Wintzen certainly worked well for his company.
Of course, the “flat” organization and a structure that preserves small-team, entrepreneurial attitudes throughout the organization is conceptually similar to the management styles found in many other lean, agile organizations. So, perhaps Eckart’s most unique contribution to management theory was the vivid simile of cellular division. That doesn’t necessarily diminish him or his ideas; after all, the ability to express a vision that’s easily understood and shared is a key skill in a founder!
As I noted off the top of this post, the “problem” of maintaining our culture while rapidly scaling up is a good one to have. I prefer thinking about our company as a growing collection of cells rather than describing it in institutional terms like “departments” but no matter how it’s worded, our vision for Weaviate has always been a flat organization made up of close-knit teams that maintain a start-up vibe.
Weaviate is open source, and you can follow the project on GitHub. Don’t forget to give us a ⭐️ while you are there!