By Dan Greenberg
Planning and strategy should take incrementalism and history into account
Sam Altman and his associates reportedly have a betting pool about when there will be a $1 billion company staffed by a single person. The premise, of course, is that AI agents can carry out all the other functions that companies would ordinarily need, so the company itself can be run by a singular human being who has created or purchased custom agents to complete said functions.
This has obvious and wide-ranging implications for business leaders around planning and strategy. Should business leaders be thinking about new sets of business tools and vendors? Should business leaders be thinking about new operational models? Should business leaders be thinking about new organization charts? And maybe most obviously, and starkly, how should business leaders be thinking about their current staffing needs and capacity?
Sam Altman is a genius and far be it for me to argue with him. On this point, however, he is wrong, and here’s why. Sam (Yes, I’m on a first name basis with him even though he does not know who I am) is looking at companies in terms of how we define and understand a company today. However, as soon as the development of AI tools allows people and companies to automate today’s functions and turn their focus to new functions, ideas, and ways to service customers, people will optimize themselves and companies will look very different than they do today.
Let’s look at this another way. Human Resources does not have a long history; it’s existed for a much shorter time than you think. But over time, companies started to make more money, and people began to find new ways to exploit companies. Companies became more efficient through automation and technology, and therefore more protective of their money, property and the status quo. They also had more money around because of said efficiency, and built HR teams to protect themselves legally.
Of course, HR teams have evolved to do more than that, but the basic function of an HR team is to protect a company, legally, and the need for that functionality only exists because it can exist. In other words, greater efficiency has led to the ability to add functions that are now necessities, but would have been seen as luxuries in the past.
It’s similar to Parkinson’s Law, which states “Work expands so as to fill the time available for its completion.” In his 1955 essay, historian C. Northcote Parkinson also talked about the idea that the number of workers within public administration, bureaucracy or officialdom tends to grow, regardless of the amount of work to be done. This he attributed mainly to two factors: that officials want subordinates, not rivals, and that officials make work for each other. As our corporate environments continue to look and feel more bureaucratic, we can clearly see the crossover between the public domain discussed by Parkinson, and the private corporate domain.
I gave the example of human resources above, but let’s also look at customer success teams. The customer success function, like HR, did not exist in the not-too-distant past. Organizations had product, sales and administration departments/areas, and sellers dealt with clients, so there was no need for a dedicated client success or service function, specifically in the B2B world.
Over time, as markets became more congested and fragmented, and switching costs decreased, it became easier for clients to churn, so it made more sense for companies to start hiring more people and spending money to keep clients around. As the technology fueled competition, and made room for more slices of the pie to be cut, it also grew the size of the overall pie, and therefore freed up money for businesses to focus more on interpersonal interaction and relationships. In other words, more technology made it so that companies could find a competitive advantage in spending more time on the non-technical pieces of their business in order to differentiate.
We can argue about whether this is good or bad for business, or for clients, or for the economy, but the point is that, historically, step function leaps in technology have not led to smaller workforces. In fact, they have led to larger workforces once the market has adjusted and companies have figured out what to compete on. Sure, Sam may be technically correct in a kind of transition period, as there may be companies who find efficiencies before the overall market adjusts. Once we reach economic equilibrium, or anything close to it, companies will find something to compete on, or something to leverage for efficiency, and will therefore not want to cut resources. It will be more efficient and profitable to add resources to improve their abilities in these areas, and that will include adding people.
Let’s look at another historical example for fun: spreadsheets. Electronic spreadsheets vastly diminished the amount of time that an accountant needed to spend building reports. Could you imagine writing numbers into graph paper every time you needed to build a revenue model? However, the advent of Excel and Google Sheets did not lead to smaller accounting teams, it led to more historical modeling, and the need for higher levels of certainty and greater reporting detail which has led to larger accounting and finance teams, and the proliferation of forward-looking modeling and financial analysis. Reporting that was unimaginable 40 years ago has become table stakes for any business.
“I get it,” Sam might say, “but AI is different because it can do anything that humans optimize themselves to be able to do.” In other words, we no longer need an accountant to run that report, because an AI agent can do it. So we are in a fundamentally different place in terms of technology development. He has a point here, but let’s clarify.
As it stands now, we don’t have Artificial General Intelligence (AGI) and we won’t have it until mid-century at the very earliest (according to the most recent predictions). AGI will conceptually be able to ingest wide arrays of inputs and act with human levels of expertise across many areas of function as opposed to what we have now which is limited by task and breadth of function.
But even as we develop AGI, assuming it plays out as we expect it to, AI already is, and will always be better than humans at many things and not as good as humans at other things. The most successful companies will know how to leverage comparative advantage. AI models will have to compete with other AI models and there is a significant cost to customize and host, so humans are likely to retain a comparative advantage in the areas where we are more efficient, like human interactions, judgement and planning.
Think about it: as AI models continue to improve, and are forced to compete against each other, we will need humans to build those ever-improving models, so at a certain point, it will make more sense to find the jobs that humans do best and and focus the AI on the things it does best rather than continually iterating on new AI models for each and every function. This is what mature markets will bring to AI development, when we get there.
We can also look at this from a very human perspective. Fads tend to change over time to reflect what is hard to acquire. We see this with the retro bespoke style movement. We can also see this, interestingly, through sociological studies on skin color treatments. In most northern European and North American countries, historically, tanning and skin darkening has been a sign of beauty, while in tropical countries it has been much more likely to be the case that skin lightening treatments have been signs of beauty. Simply put, things that are harder to come by are seen as valuable for the very reason that they are harder to come by.
So, it is very likely that over time, as it becomes increasingly easier to execute on functions and outputs that differentiate businesses today, it will end up being the things that businesses cannot do today, that new technology will eventually allow them to do, that will be competed on most fiercely. Those new functions and outputs will be harder to execute on, and therefore more valuable to customers. Those new functions and outputs will be at the cutting edge of technology, and therefore necessarily need human input and humans to work alongside the technology. And because those new functions and outputs will be harder to compete on, businesses will look for ways to gain an advantage. And that will often mean hiring more people to perform in very different-looking roles, but nonetheless, hiring people.
This will mean that in the not-to-distant future all the functions that we currently understand that need to be accounted for in a company may be able to be executed on by one person and a cadre of AI agents. However, the new frontier in business will be the functions and outputs that differentiate one business from another in customer service, in efficiency, and in product-to-market capabilities. These things will always exist at the vanguard of technological improvement, therefore always necessitating some level of human involvement and teamwork between humans and technology.
So, what does this all mean for planning and strategy decisions in the coming years? We don’t know exactly what the world will look like once AI tools are more ubiquitous, but it seems a safe bet to assume that it will be very important to focus resources on how to build productive relationships between AI tools and highly skilled workers. In many cases, this will mean investing in people and processes that focus on how to input information into tools, and how to extract information and usable outputs from those tools that can be leveraged by those highly skilled workers in order to make themselves more efficient.
Thinking about the problem from this angle can help business leaders think about tools, operational models, and organizational charts in a way that supports the world as we know it and the potential future world that is still unclear and evolving.
___________________________________________________________________________________
Canopy Expert Advisor Dan Greenberg is a revenue leader with general management background and an outstanding track record of growing revenue organizations through thought leadership and operational rigor, as well as developing internal and external relationships and partnerships.
Interested in bringing top-tier fractional experts like Dan into your organization? Tell us about your project here.
