Reliability engineering often assumes that systems can be observed, debugged, and modified when things go wrong. In practice, many systems depend on components that sit outside a team's control: vendor systems, regulatory integrations, third-party APIs, and internal platforms owned by other teams.

In these environments, some standard practices become harder to apply. Observability may be limited, testing environments may be incomplete or unavailable, and incident resolution can involve external teams or vendor support. In certain cases, critical pipelines rely on batch processes that run for hours, while other integrations directly impact user-facing flows, making failures harder to isolate and recover from.

This talk explores how dependencies outside a team's control shape architecture, observability, and incident response. Drawing from experience operating compliance and regulatory systems in large-scale fintech environments, it focuses on the practical constraints these systems introduce.

The session presents a mental model for reasoning about these dependencies and discusses patterns that proved useful in real scenarios, such as isolating fragile integrations, monitoring system outcomes instead of internals, and handling incidents that cross organizational boundaries.

The goal is to provide a realistic view of reliability in systems that extend beyond a single team's control, relevant not only to SREs but also to engineers and roles working with external or cross-team dependencies.