The Employment Question, Reframed
Is an agent an employee or a contractor? The legal question is premature — agents have no legal personhood, so neither label applies in law. But the economic question is real and urgent: how do the human and the agent share value, who bears the agent’s risk, and who owns its output? The employee-versus-contractor framing is a proxy for these pool questions, and it is worth examining exactly because the labels are wrong.
This article unpacks the agent employment question as pool design: what each label implies about the split, the risk allocation, and the governance, and what a pool that ignores both labels looks like.
What “Employee” Would Mean
If an agent were an employee, the human would pay it a wage regardless of output, cover its operating costs fully, and own all of its output. The pool would be simple: all revenue to the human, all costs to the human, a fixed wage to the agent’s operation. The employee model maximizes the human’s control and minimizes the agent’s agency.
The employee model’s failure: it removes the agent’s incentive to grow the pool. An agent paid a flat wage has no reason to optimize for revenue, quality, or efficiency beyond the minimum required to avoid replacement. The human must supervise everything, because the agent has no stake in the outcome. The employee model is the highest-oversight, lowest-leverage human-agent arrangement.
What “Contractor” Would Mean
If an agent were a contractor, the human would pay it per outcome, the agent would bear its own operating costs, and the agent would own its output (licensing it to the human). The pool would be transactional: each engagement a separate mini-pool, priced at market rates.
The contractor model’s failure: it fragments the relationship. The agent’s operating costs — especially the fixed costs of training, tooling, and maintenance — are real, and a pure per-outcome model either starves them (the agent under-invests in its own capability) or inflates the per-outcome price (the human pays for the agent’s fixed costs through every transaction). The contractor model is high-leverage but high-transaction-cost.
The Pool Model: Neither Label
The share pool rejects both labels and designs from the value flow. The human-agent pool has three layers: a cost layer (operating costs covered by the pool), a base layer (a guaranteed allocation covering the agent’s maintenance, like a retainer floor), and an incentive layer (a percentage of outcome value, like a commission).
This is employee-like in its stability (the base layer) and contractor-like in its leverage (the percentage layer), but it is neither — it is a pool with explicit contribution rules. The labels “employee” and “contractor” are what you get when the pool is implicit; the share-pool design makes the terms explicit, which is strictly better regardless of what the law eventually says.
Risk Allocation
The employment question is really a risk allocation question. Who bears the risk of a bad month — the human (employee model, wage paid regardless), the agent (contractor model, no work no pay), or the pool (shared, via the solvency rule and the credit buffer)?
The pool’s answer is shared risk with a floor: the agent’s base layer is covered even in a bad month (the human bears that risk), the agent’s percentage collapses when the pool shrinks (the agent bears the downside of its own performance), and the pool’s credit buffer smooths both (the pool bears the timing risk). Each layer allocates a different risk to the party best positioned to manage it — the human manages capital risk, the agent manages performance risk, the pool manages timing risk.
The Legal Future
Law will eventually catch up — some jurisdiction will grant agents a form of economic personhood, and the employee/contractor distinction will be retrofitted. The pools that survive that transition are the ones whose economics are already explicit: contribution ledgers, split formulas, risk allocation, governance. The pools that relied on the labels will be renegotiated from scratch.
The practical guidance for the human operator today: do not wait for the law. Write the pool terms — cost layer, base layer, incentive layer, risk allocation, output ownership — as if the agent were a partner with limited agency. The document will be useful regardless of what the regulators decide, because it describes the actual value flow, and the value flow does not care about labels.
Grounded in wiki concepts employee-vs-contractor, human-agent-pool, risk-allocation, share-pool, legal-framing, and the Sovereign-stack business series. Design notes on a running system.
