If Telegram ingress lives inside the same process as the orchestrator, teams do not get automation. They get another blind spot. Before this change, the path from Telegram webhook to AgentSyncHub intent routing was too easy to treat as "part of the app". When that path degraded, the failure signal was muddy: bot silence, mixed logs, shared process state, no separate readiness boundary. Now the ingress runs as its own telegram-bot process with its own /ready and /v1/telegram/health, and it hands intents to AgentSyncHub over explicit HTTP with auth instead of hidden in-process calls. That changes the operational model. Message intake, bot health, and workflow submission are now observable as a chain instead of a guess. A healthy backend no longer masks a dead chat entrypoint, and a chat issue no longer looks like a generic platform wobble. For CTOs, Staff engineers, and platform teams, this matters because most AI automation does not fail at the model layer. It fails at the boundary