The Company of One: How Agentic AI Will Reinvent Business Development

There is a question I have been sitting with for the past year that I think every founder, every executive, and every person building something in the world right now should be sitting with too.

What does a company actually need humans for?

Not as a provocation. As a genuine, honest question. Because the answer is changing faster than most organisations have begun to appreciate, and the implications of that change extend far beyond efficiency gains or headcount decisions. We are moving toward a fundamental redefinition of what a company is, how it comes into existence, how it operates, and what role human beings play within it. The catalyst is agentic AI, and the trajectory, if you follow it honestly to its logical conclusions, is one of the most extraordinary shifts in the history of organised human activity.

I want to try to describe what I see coming, what it means in practical terms for founders and builders today, and how I think about positioning ourselves to benefit from it rather than be overtaken by it.

From Tools to Agents

To understand where we are going, it helps to be precise about where we are.

For the past few years, most people have experienced AI as a tool. You open an interface, you ask a question, you receive an output. The AI responds to your prompt and then waits. It has no memory of yesterday, no awareness of tomorrow, and no ability to take action in the world beyond the text it produces. It is extraordinarily capable within those constraints, and it has already changed how most knowledge workers do their jobs. But it is still, fundamentally, a very powerful typewriter.

Agentic AI is something categorically different. An AI agent does not just respond. It plans, executes, monitors, adapts, and loops back. It can be given a goal rather than a task, and it will work out the steps required to achieve that goal, call the tools it needs, process the results, adjust its approach when something does not work, and continue until the objective is met or it determines that it cannot be. It operates over time, not just in a single exchange. And increasingly, agents can coordinate with other agents, delegating subtasks, aggregating outputs, and managing complex workflows that would previously have required an entire team of human specialists.

We are in the early stages of this transition. The agents available today are impressive but imperfect. They make mistakes, lose context, and require supervision. But the rate of improvement is steep and consistent, and the direction is not in doubt. Within the next five to ten years, agentic AI will be capable of executing the majority of the discrete professional tasks that make up the operational backbone of most businesses, not occasionally, not experimentally, but reliably and at scale.

What This Means for How Companies Are Built

Let me be concrete about what this actually changes, starting from the very beginning of the business development journey.

Today, starting a company requires assembling a team. You need people who can research the market, validate the idea, design the product, write the code, handle the finances, manage legal compliance, run customer support, produce marketing content, analyse performance data, and iterate based on what they find. Each of these functions has historically required dedicated human expertise, and the cost and complexity of assembling that expertise has been one of the primary barriers to entrepreneurship.

Agentic AI dismantles that barrier almost completely.

Imagine a founder in 2027 with an idea for a new service. Rather than spending six months assembling a founding team and raising seed capital to pay for it, she describes her vision to an orchestration layer of AI agents. One agent conducts market research, analysing competitor positioning, customer sentiment, pricing dynamics, and unmet needs. Another synthesises that research into a strategic brief. A third translates the brief into a product specification. A fourth builds a working prototype. A fifth generates the brand identity, the copy, and the initial marketing materials. A sixth sets up the legal structure, the payment infrastructure, and the compliance framework. A seventh runs the first customer acquisition experiments and reports back on what is working.

All of this happens in days, not months. At a fraction of the cost. With the founder's role shifting from doing to directing, from executing to deciding, from managing people to managing agents.

This is not science fiction. The component parts of this workflow exist today in early form. The integration layer that makes it seamless is what the next few years of AI development will produce. And when it arrives, the implication is profound: the cost of starting a company will approach zero, and the speed at which a validated, operational business can be assembled will compress from years to weeks.

The Abstraction Layer Keeps Rising

Here is what I find most striking about the trajectory of agentic AI, and it is something that I think most current conversations about AI underestimate.

The abstraction level at which humans interact with AI systems will keep rising. This is a pattern that has repeated across every major wave of technology, and it is accelerating now.

In the early days of computing, you had to write machine code. Then programming languages abstracted that away. Then integrated development environments abstracted the languages. Then no-code tools abstracted the environments. At each step, the human moved further from the underlying complexity and closer to the level of intent, focusing more on what they wanted to achieve and less on how to achieve it.

AI is doing the same thing, but faster and across a far wider domain than software development. We are moving from prompting, which is still a relatively low-level interaction where you have to be specific and precise about what you want, toward intention, where you describe an outcome and the system figures out the rest. And from there we will move toward autonomy, where AI systems pursue long-horizon goals with minimal human input, checking in only at decision points that require human judgment or approval.

What this means practically is that the competitive advantage in business will shift from execution to vision. The ability to do things, which AI will increasingly handle, matters less. The ability to know what is worth doing, to see an opportunity clearly, to define a goal that is genuinely valuable and articulate it in a way that can be pursued, that becomes the scarce and valuable input.

The founder who can describe what the world needs, clearly and specifically enough that an AI system can move toward it, is extraordinarily powerful. The professional who can only execute what someone else has already envisioned is in a much more precarious position.

Higher Autonomy and What It Demands of Us

As AI agents become more autonomous, the nature of human oversight will change in ways that are both liberating and demanding.

The liberating part is obvious. Tasks that currently consume enormous human time and energy will simply happen. Operational management, reporting, customer communication at scale, content production, financial monitoring, compliance checks, competitive intelligence gathering. All of this will increasingly run in the background, supervised but not driven by human effort.

The demanding part is less often discussed. When AI systems are making thousands of decisions on your behalf every day, the quality of the instructions you give them, the values you embed in them, and the judgment you apply when reviewing their outputs becomes critically important. The leverage is enormous, which means the cost of a poorly defined goal or a misaligned value is also enormous.

This is one of the reasons I believe that the human skills that will matter most in an agentic AI world are not technical ones. They are the skills of clarity, judgment, and accountability. The ability to define what you actually want, to distinguish between a good outcome and a plausible-sounding but wrong one, and to take responsibility for the decisions that AI executes on your behalf. These are fundamentally human capacities, and they will be in higher demand, not lower, as autonomy increases.

There is also a deeper question here about trust. As AI agents take on more consequential roles in business, the question of how much autonomy to grant them, in which domains and under which conditions, will become one of the central governance challenges of our era. The organisations that get this calibration right, that find the balance between human oversight and AI execution that maximises outcomes while managing risk, will have a significant advantage over those that either over-delegate or under-utilise.

What Business Development Looks Like at the Next Level

Looking further out, perhaps ten to twenty years, I think we will see the emergence of something genuinely new: businesses that are largely conceived, built, and operated by AI systems, with humans involved at the level of ownership, governance, and strategic direction rather than day-to-day management.

The business development process will be transformed at every stage. Ideation will be augmented by AI systems that can scan markets, identify patterns, simulate customer responses, and surface opportunities that no individual human could have spotted. Validation will happen faster, with AI agents running experiments, gathering data, and synthesising conclusions in real time. Product development will be continuous and automated, with AI systems iterating on design and functionality based on live user behaviour without waiting for quarterly planning cycles. Go-to-market will be personalised at a level that current marketing teams can only dream of, with individual messages, offers, and experiences generated dynamically for each potential customer. Operations will be self-managing, with AI monitoring performance, identifying bottlenecks, and resolving them without human intervention in the vast majority of cases.

What remains distinctly human in this picture is not nothing. It is, in fact, the most important part. Setting the direction. Defining the values. Making the calls that require wisdom and experience rather than information processing. Building the relationships, partnerships, and trust that determine whether a business is a participant in a healthy ecosystem or an extractive actor in a broken one. Deciding what is worth building in the first place, and why.

These are not trivial residuals. They are the things that have always mattered most in business and that have historically been crowded out by the sheer volume of operational work surrounding them. Agentic AI, at its best, clears away that operational noise and puts these essential human contributions back at the centre.

How I Think About Positioning Yourself for This

I have spent a lot of time thinking about this question, both as a founder navigating the transition in real time and as someone who thinks carefully about what it means to build something durable.

The frame I keep returning to is this: the people who thrive in an agentic AI world will be the ones who learn to think at a higher level of abstraction than the people around them. Not because abstraction is inherently virtuous, but because the ability to move between levels, to understand the detail well enough to direct it meaningfully, and to see the whole clearly enough to make good strategic choices, is exactly what AI cannot do and what organisations will need most desperately.

Concretely, that means getting genuinely fluent with AI tools now, not as a curiosity but as a working practice. It means developing the habit of asking what you are actually trying to achieve before you ask how to achieve it. It means investing in the judgment, taste and values that will determine whether the outputs of increasingly autonomous AI systems are ones you are proud of. And it means staying curious about the direction the technology is moving, not to predict it precisely, which nobody can do, but to remain oriented and adaptive as it evolves.

For founders specifically, I think the opportunity is extraordinary and underappreciated. The cost of building is collapsing. The speed of iteration is accelerating. The ability of a small, sharp team with excellent AI tooling to compete with organisations ten times their size is increasing every month. The founder who combines genuine domain insight with AI fluency and the human skills to lead through complexity is in perhaps the strongest position in the history of entrepreneurship.

The companies of the future will not look like the companies of the past. They will be leaner, faster, more global, and more capable. The humans within them will be fewer in number and higher in impact. The tools available to them will be extraordinary.

The question, as it has always been, is what you choose to build with them.

I think that is worth getting very clear about.