APIs Replaced Managers. AI Brings Them Back.

In 2015, Peter Reinhardt warned that APIs were quietly replacing middle management. Those “above the API” (engineers, designers, knowledge workers) would thrive, while those “below it” (drivers, delivery workers, support agents) would face automation. He was right for his time. Ten years later, the rise of AI shows why that picture is more complex.

When Non-Determinism Meets Determinism

APIs are deterministic. Ask for the status of a payment, and you always get the same answer. That predictability made APIs powerful but also brittle. They are powerful because you can count on them to behave the same way every time. They are brittle because they cannot adapt to unexpected events. If the wrong input arrives, if calls stack in the wrong order, or if an external system changes the context, the API does not bend. It breaks. Once work is abstracted into endpoints, it becomes exposed to whoever or whatever calls them.

AI is different; it is non-deterministic: it interprets, adapts, and improvises. Ask it to explain a contract, and you may get a different phrasing each time. APIs bring stability, while AI brings variability. When combined, they generate volatility.

Volatility is not inherently bad. It is the raw material of adaptation, just as variation in nature is the raw material of evolution. Without it, systems cannot learn or adjust. However, without stabilizers, minor variations ripple through pipelines built for reliability, and what appears to be smooth automation can drift or collapse.

This is where orchestration and judgment matter. Orchestration channels variability into structured flows that enterprises can manage. Judgment ensures that when surprises arise, humans determine their meaning and decide on the next course of action. Together, they transform volatility from a risk into a source of resilience.

From APIs to Orchestration and Judgment

The orchestration graph is the living blueprint of how work flows across humans, agents, and systems. It makes clear who does what, in what order, and with what tools. Without orchestration, variability turns into noise. With it, it becomes a manageable flow.

The judgment graph shows where humans step back in. These are the moments when rules, policies, or automated responses are not enough. Judgment restores context. It turns exceptions into decisions and ensures that meaning is not lost in translation.

On their own, workflows are rigid. They prescribe order and structure. But when designed to include points for human judgment, they act as stabilizers. The workflow provides the scaffold, and humans supply the context. Together, they absorb volatility, catch exceptions, and keep the system aligned with values and intent.

APIs erase variation, AI introduces it. Orchestration and judgment give that variation shape and meaning. What looks inefficient on the surface is, in fact, the foundation of resilience.

The Risk

Efficiency hides its own risks. An enterprise that optimizes only for speed and cost looks smooth on the surface, but beneath it, every exception compounds. The result is not just technical fragility but institutional risk: decisions no one can explain, values stripped from outcomes, and trust eroded at scale. The danger is not collapse alone, but collapse that arrives unnoticed until it has already spread. Without judgment built in, efficiency becomes a liability rather than an advantage.

The Opportunity

The risks are real, but they also force a better design. Efficiency alone produces fragility, which means enterprises can no longer ignore oversight, values, and judgment. AI’s unpredictability makes these elements non-negotiable.

This is the source of optimism. The very volatility that can break systems also creates the possibility of more resilient ones. Workflows once dismissed as inefficiency become the features that keep enterprises adaptive and alive.

APIs stripped away middle management. Orchestration and judgment put us back where we matter most: not as bottlenecks but as governors of resilience and meaning. Everyone becomes a manager of agents, decisions, and values.

The future of the enterprise will be defined not by efficiency alone, but by the quality of its orchestration and judgment - turning volatility into strength.

Thanks to Jules Qin for pointing me to the original post