Earlier in my career, I worked as a management consultant doing organizational (re)design projects for retailers and CPG companies.
We always told the companies we worked with:
- Start with ‘design principles’ anchored on commercial objectives
- Draw up the ideal “blank sheet” org structure, then;
- Layer in practicalities and constraints
Unsurprisingly (and frustratingly) the practical constraints often neutered commercial design principles and overwhelmed good first principles design.
I now find myself pulling out that toolkit as we build Beacon’s central organization, and those of our portfolio companies.
Doing so in a world where the very nature of white collar work is changing so rapidly with the daily advances in AI presents unique opportunities and challenges.
The topic of organization design in an AI world also seems to be a recurring theme not only with startup peers but also with CEOs of scaled growth technology companies and real world/main street businesses.
In a quest for feedback/pushback and also to try to be helpful I wanted to share how I am thinking about org design in the age of AI.
Design Principles:
- Speed of decision making and execution: We will look back fondly on the days where we had the luxury of creating buy-in, building consensus, shaping a decision via multiple meetings stretched out over weeks (or months) and then rolling out decisions over multiple quarters. As AI reduces barriers to competitive entry and increases customer expectations, survival will require ‘big’ decisions to be made quickly, and changes to be shipped immediately. In a world where agents can coordinate tasks amongst themselves instantly and 24/7, the market will expect an instant problem → response loop. Organizations need to be designed to allow for that to happen.
- Minimize layers: So much of traditional org design has been about balancing between managerial spans-of-control, and the number of layers between executives and the lowest level of an organization. Spans-of-control become irrelevant in a world where executive bandwidth grows exponentially with AI tools and where many roles will be filled by AI agents instead of humans. Middle layers that exist to gate, summarize or cascade information will be irrelevant as every environmental input is converted into a piece of data, which is subsequently synthesised and acted on by PhD level AI agents.
Implications:
- End of the ‘Coach’ Only Role: Every leader will have to be a player first and coach second. When the content and nature of work are changing so quickly, the only way to be an effective coach of others is to know the work yourself.
- The Death of Functions: In an environment where any question, regardless of how complex, is answerable with a $10/month subscription humans will no longer be defined by technical specialization. ChatGPT can explain the nuances of Delaware C-corp law or the principles of API authentication. When knowledge is commoditized, roles built solely on specialization become irrelevant.
- Impermanence: I am convinced no one really knows what the next five years will look like as AI invades our lives. As a result, organizational structures will be built and destroyed quarter-over-quarter if not month-over-month. Winning companies will seek to maximize flexibility. Winning employees will get comfortable with transient relationships with ‘roles’ (and sometimes employers).
- CEO of Control Center: The modern CEO will be less of a strategist or cheerleader, and more of an F1 driver. They will spend their days interpreting real-time telemetry, responding instantly, and orchestrating both humans and machines. It is a shift from planning to navigating. From delegation to direct action.
So what?
If you are building or retooling your organization in this environment:
- Treat your structure as temporary
- Redesign roles around outcomes not functions
- Hire leaders who do work, not just manage or inspire
- Decide if you want to be in the cockpit of your race car