"Process Zero"Thinking
Alena stands. The room is silent. She looks at each board member in turn. "How do you architect for a world that already thinks faster than you do?"
She pauses. "The natural instinct is to ask 'how do we transform from where we are today?' That's the wrong starting point. Working from today's processes will limit your vision inside the first half of the chessboard."
The fundamental question
Instead, leaders should work backward from the compounded future: what must be true and what must remain human for our business to thrive when intelligence is purchasable by the token?
The shift in computing
$0M
Average enterprise LLM spend in 2025
Up from $4.5M in 2024 -- overshooting projections by 56%
$2K-$20K
OpenAI's planned AI employee tiers per month
Knowledge workers to research agents
0%
Agentic AI projects Gartner warns will be cancelled by 2027
Due to infrastructure unreadiness, not model limitations
What Must Be True for "One-Person IT" to Succeed?
A reliable AI and agentic partner ecosystem
The architecture depends on a mature, interoperable ecosystem of AI platforms, foundation models, and agent protocols (Agent skills, MCP, A2A) that can be composed like building blocks. The "one person" isn't doing the work -- they're orchestrating an ecosystem that does.
Enterprise-grade infrastructure that runs itself
Cloud-native, API-first infrastructure with self-correcting, auto-scaling, and built-in observability must underpin every layer. Gartner warns that 40% of agentic AI projects will be cancelled by 2027 due to infrastructure unreadiness.
Sovereign data architecture from day one
The true moat is the proprietary feedback loop. That loop only works if data flows freely across the enterprise in your schema, on your terms -- not locked inside vendor platforms.
Operate in the token economy
The unit of work is now the token: a unit of purchased intelligence. The one-person IT model must be built around token economics as a core competency -- routing the right task to the right model at the right cost.
Governance baked into architecture
Policy-as-code: your regulatory and ethical rules written as executable logic, version-controlled alongside your application code, and enforced automatically. "No agent can access PII without role-based clearance." "No deployment passes without encryption."
Trust from boardroom, customers, and regulators
Two-thirds of board members admit limited AI knowledge. 76% of consumers would switch to a brand offering genuine AI transparency. The societal permission to operate autonomous systems is being written right now.
What Can't "One-Person IT" Achieve?
Cross-organisational trust and stakeholder relationships
AI can prepare the analysis, but building trust with regulators, partners, and employees requires human presence and political judgement.
Single point of failure risk
One person controlling all knowledge and decision authority is the textbook definition of a single point of failure -- the very risk enterprise architecture exists to eliminate.
Genuine ambiguity and novel crisis
Unstructured, novel, high-stakes crises require contextual judgement, moral reasoning, and adaptive improvisation that remains uniquely human.
The creative leap that defines strategy
AI cannot determine which questions are worth asking in the first place. Defining what to build and why it matters still requires a human mind with lived context.
The physical world
Data centres still need electricians. Hardware still fails. The digital layer can be orchestrated by one person, but the atoms underneath remain stubbornly human.
The Pathway Every CTO Must Consider
No one can predict what enterprise IT will look like in 2030. These are six no-regret decisions that increase the organisation's degrees of freedom as AI capabilities compound.
The Closing Statement
"The 'one-person IT' model is not a fantasy of a large enterprise run by a single technologist. It is a redesign of the enterprise IT function so that coordination, execution, and verification can be carried by software agents -- with a smaller number of humans owning intent, accountability, and the moments where trust matters."
"None of these moves require you to predict the future. Each makes the enterprise stronger whichever future arrives. Together, they move IT from 'keeping the lights on' to shaping what the business can safely attempt, how fast it can learn, and how confidently it can operate when autonomy becomes normal."
The chessboard is already in play.
The question is which square you're on, and whether you've started compounding.
The pace is already high. The future will stay uncertain.
What is not uncertain is the cost of waiting. It compounds quietly, then all at once.
By Jack Wang, Reinvention Worx -- February 2026
The Future of Enterprise IT