Self-Service Provisioning
A developer obtaining infrastructure through an automated interface bound by policy, with no ticket and no human approval in the path.
The economics are the whole argument. A ticket-based model has a fixed human cost per request and a queue whose length grows with the organisation, so the infrastructure team becomes a bottleneck precisely as the company scales. Self-service has a fixed build cost and near-zero marginal cost, which is the only shape that survives growth.
The objection is always control, and the answer is that control moves rather than disappears. The constraints are encoded in what the interface can produce: a database request yields an instance in an approved size range, in an approved region, encrypted, backed up, tagged to a cost centre, network-isolated by default. The developer is choosing inside a space the platform has already made safe, which is stronger control than a reviewer approving a free-form request under time pressure.
Where it fails is scope creep in the interface. Every additional option multiplies the combinations the platform must support, and a provisioning API with forty parameters has recreated the complexity it existed to hide.