Amazon Bedrock Flow
Does ZopNight manage Amazon Bedrock Flow?
Bedrock flows bill through their nodes: each execution charges for the model invocations, agent runs, and knowledge-base queries the flow chains together, so run volume times per-run cost is the whole bill. ZopNight discovers flows via its Bedrock provider on the 6-hour cycle and trends execution-driven cost from Cost Explorer or CUR 2.0.
Rules that fire on Amazon Bedrock Flow
No active rule family targets Amazon Bedrock Flow today. Rules that used to are retired, and retired rules publish no pages and fire no findings. Scheduling and permissions coverage are unaffected.
At a glance
| Field | Value |
|---|---|
| Scheduling notes | discovery and cost tracking only. |
A Bedrock flow is a visual workflow that chains prompts, models, agents, and business logic. Flows generate cost through the model and service calls in each execution, so run volume drives spend.
Per-execution economics
A flow’s definition is free to store; its executions are where money moves. Each run walks the graph, and every node that touches a billable service meters normally: prompt nodes bill their model’s input and output tokens, agent nodes bill their full reasoning loops, knowledge-base nodes bill retrieval, Lambda nodes bill compute. The flow’s total is the sum of its path, which makes per-run cost a property of the graph’s design: how many model nodes a typical execution visits, which models they call, and how much text each passes forward. Multiply by run volume and that is the entire bill.
Flows as cost units
ZopNight’s Bedrock provider discovers flows on the 6-hour cycle and trends their execution-driven cost from Cost Explorer or CUR 2.0. Treating the flow as the unit of analysis matches how spend actually concentrates: individual model calls are anonymous, but a flow is a named business process (the document summarizer, the support triage pipeline) whose cost per period can be watched, compared, and questioned. Volume-driven growth (the flow is succeeding) reads differently from per-run growth (the flow got more expensive to execute), and the trend separates them.
Graph designs that overbill
Chained maximalism is the common flaw: a flow that routes every input through its most capable model when a classifier node could send easy cases down a cheaper path. Redundant context passing bills the same document’s tokens at node after node instead of extracting once. Retry wiring without limits turns transient downstream errors into multiplied model calls. And scheduled flows keep executing on cadence after their consumers churn, orchestrated diligence for an audience of zero.
Reviewing flow definitions
The Bedrock console’s Flows view lists each flow with its node graph and version history. Walking the graph with per-node model choices in view is the cost review: every model node is a rate decision, and every edge is a chance to send fewer tokens.