Amazon Bedrock Agents Monitoring

Amazon Bedrock Agents - Overview

Amazon Bedrock Agents is a feature of Amazon Bedrock that enables developers to build autonomous AI agents capable of executing multi-step tasks by orchestrating foundation models, APIs, and knowledge bases. Agents interpret natural language instructions, break down complex requests into sequences of actions, invoke appropriate tools (such as Lambda functions and knowledge bases), and return coherent responses - without requiring custom orchestration logic.

Monitoring Amazon Bedrock Agents is critical for ensuring reliable AI workflow execution, controlling costs, and maintaining service performance. Applications Manager's Amazon Bedrock Agents monitoring tool provides comprehensive visibility into key parameters such as invocation success rates, total response time, time to first token, model latency, throttling events, and token consumption across each agent alias and operation. With real-time alerts, historical performance trends, and per-alias breakdowns, teams can quickly detect agent failures, optimize model usage, and ensure that AI-driven workflows meet performance and cost targets.

Creating a new Amazon Bedrock Agents monitor

To learn how to create a new Amazon Bedrock Agents monitor, refer here.

Monitored Parameters

Go to the Monitors Category View by clicking the Monitors tab. Click on the Bedrock Agents instance available under Amazon in the Cloud Apps section. Displayed below is the Amazon Bedrock Agents bulk configuration view, distributed into three tabs:

Availability tab gives the availability history for the past 24 hours or 30 days.
Performance tab gives the health status and events for the past 24 hours or 30 days.
List view tab enables you to perform bulk admin configurations.

By clicking a monitor from the list, you'll be taken to the Amazon Bedrock Agents dashboard, which includes the following tabs:

Configuration Overview

Parameter	Description
AGENT INFORMATION
Lifecycle State	The operational availability status of the designated Bedrock Agent at the time of polling.
Description	The user description details are configured for the agent.
Agent Version	The current active semantic deployment version identifier of the agent at the time of polling.
Foundation Model (LLM)	The target large language model (LLM) blueprint is configured as the core reasoning engine for this agent instance at the time of polling.
Creation Date	The creation timestamp indicates when this functional asset was originally launched inside the region.
Last Updated	The revision timestamp indicates when this profile's rules or limits were last updated.
AGENT CONFIGURATION
Guardrail ID	The identifier of the active guardrail associated with this operational agent at the time of polling.
Guardrail Version	The exact structural version of the guardrail profile is coupled to this agent lifecycle.
Orchestration Type	The core reasoning framework layout (such as Zero-Shot React or structured step-by-step planning) drives the agent's multi-turn task execution loops at the time of polling.
Memory Type	The list of active conversation memory persistence frameworks configured to retain rolling user context windows across sessions at the time of polling.
IAM Role	The AWS Identity and Access Management (IAM) execution role name assigned to this agent at the time of polling defining its backend resource interaction boundaries and cross-service permissions.
Session Idle Time-to-Live	The session validation Time-to-Live (TTL) configuration at the time of polling, marking the maximum idle duration before a user's conversation state context is automatically recycled (in seconds).
Table: Version Details
Version ID	The numerical or descriptive build identifier indicating this specific immutable snapshot version of the agent at the time of polling.
Description	The historical release description statements are logged when freeze locking this specific agent build version at the time of polling.
Creation Time	The precise timestamp recording when this immutable release configuration version was built at the time of polling.
Last Updation Time	The timestamp tracking the last update or metadata refresh applied across this configuration version context at the time of polling.
Build State	The system availability and execution health state tracking of this frozen agent build profile at the time of polling.
Table: Version Association Details
Version ID	The numerical or descriptive build identifier indicating this specific immutable snapshot version of the agent at the time of polling.
Linked Guardrail ID	The identifier of the active guardrail associated with this operational agent at the time of polling.
Guardrail Version	The exact structural version of the guardrail profile is coupled to this agent lifecycle.
Linked Knowledge Base ID	The distinct identifier tracking this vector storage integration.
KB Description	The architectural statement documenting this vector resource's indexing purpose.
KB Last Updation Time	The timestamp showing the last structural data synchronization or configuration commit on this Knowledge Base.
Grounding Pipeline State	The baseline synchronization or provisioning readiness phase of the Knowledge Base at the time of polling.
Build State	The system availability and execution health state tracking of this frozen agent build profile at the time of polling.

Note: Version & Knowledge Bases data collection is mapped under Performance Polling and disabled by default. To enable or disable data collection, navigate to Settings > Performance Polling. In the Optimize Data Collection tab, select Amazon Bedrock Agents as the Monitor Type, choose Version & Knowledge Bases as the Metric Name, and update the Default Polling Status as required.

Deployments & Aliases

Parameter	Description
Table: Alias Deployment Records
Alias Deployment ID	The distinct logical identifier assigned by AWS to target this specific agent deployment path at the time of polling.
Alias Name	The administrative tracking label was configured for this specific agent alias instance at the time of polling.
Description	The description is configured for this alias deployment.
Creation Time	The date when this alias deployment was created.
Agent Operation	The specific execution or inference operation identifier discovered via CloudWatch telemetry mappings at the time of polling.
Model Name	The name of the AI model used by this alias deployment.
Associated Version	The underlying semantic target version mapping driving the immutable agent snapshot targeted by this routing path at the time of polling.
Last Updation Time	The precise timestamp marking the last modification made to this resource or deployment path.
Deployment Status	The operational availability and provisioning readiness phase of this specific deployment alias at the time of polling.
Table: Orchestration Execution Performance
Alias Deployment ID	The distinct logical identifier assigned by AWS to target this specific agent deployment path at the time of polling.
Agent Operation	The specific execution or inference operation identifier discovered via CloudWatch telemetry mappings at the time of polling.
Model Name	The name of the AI model used in this invocation.
Total Requests	The total number of end-user requests routed through this deployment alias and execution operation path between the poll interval.
Client Errors	The total number of bad requests or input validation failures (4xx) received by this agent alias between the poll interval.
Server Errors	The total number of internal agent engine failures (5xx) recorded across this deployment path between the poll interval.
Throttled Requests	The total number of user requests to this specific alias path rejected or rate-limited by account concurrency controls between the poll interval.
First Token Time (TTFT)	The average Time to First Token (TTFT) streaming duration recorded for this alias route between the poll interval (in ms).
Request Latency	The average end-to-end execution latency required by the agent orchestration layer to process a user turn between the poll interval (in ms).
Prompt Tokens (Input)	The total volume of ingress prompt tokens sent to the core model configuration through this specific deployment vector between the poll interval.
Response Tokens (Output)	The total volume of egress generation tokens returned by the underlying model layer under this specific deployment vector between the poll interval.
Table: Model Invocations via Alias Context
Alias Deployment ID	The distinct logical identifier assigned by AWS to target this specific agent deployment path at the time of polling.
Agent Operation	The specific execution or inference operation identifier discovered via CloudWatch telemetry mappings at the time of polling.
Model Name	The name of the Bedrock model.
Model Client Errors	The total number of validation errors or payload structural mismatches rejected by the target model under this execution path between the poll interval.
Model Server Errors	The total number of backend system execution crashes experienced at the target foundation model layer under this routing path between the poll interval.
Model Request Throttles	The total number of rate limits and concurrent request blocks triggered directly at the foundation model tier during this alias transaction loop between the poll interval.
Model Request Latency	The average round-trip latency spent waiting on foundation model responses inside this specific execution route between the poll interval (in ms).
Model Requests	The total number of direct inference queries issued to the target foundation model version under this alias scope between the poll interval.
Top 5 Aliases by Agent Server Errors - Displays a bar chart of the top 5 agent alias deployments with the highest number of agent server errors during the selected time period.
Top 5 Aliases by Agent Client Errors - Displays a bar chart of the top 5 agent alias deployments with the highest number of agent client errors during the selected time period.
Top 5 Aliases by Model Server Errors - Displays a bar chart of the top 5 agent alias deployments with the highest number of model server errors during the selected time period.
Top 5 Aliases by Model Client Errors - Displays a bar chart of the top 5 agent alias deployments with the highest number of model client errors during the selected time period.

Amazon Bedrock Agents Monitoring

Amazon Bedrock Agents - Overview

Creating a new Amazon Bedrock Agents monitor

Monitored Parameters

Configuration Overview

Deployments & Aliases

Loved by customers all over the world

"Standout Tool With Extensive Monitoring Capabilities"

"I like Applications Manager because it helps us to detect issues present in our servers and SQL databases."

Carlos Rivero

Trusted by thousands of leading businesses globally