FlowAMP Agent Topology
36 Agents · 6 Domains · LiveRevenue Protected
$4.2M
↑ 32% vs last month
Mean Time to Intervene
14s
Anomaly → Action
Decisions / Hour
847
↑ 8% this week
Human Override Rate
1.2%
↓ 0.3% — high trust
Shutdowns Prevented
23
PSM · EPA · ISO
Cross-Agent Tasks
312
↑ 24% orchestration
Responsible AI Scores
36 AgentsRisk & Compliance Scores
10 StandardsAgent Catalog
Complete inventory of 36 intelligent agents across 6 domains
36
Total Agents
34
Deployed
6
Domains
99.8%
Availability
Total Monthly Spend
$0
↑ 3.2% MoMBudget: $24,000
Cost per 1K Invocations
$0
↓ 8% vs baselineTarget: <$0.80
Compute Utilization
0%
—Target: >80%
Projected Overrun
$0
On trackEOM forecast
30-Day Spend Trend
Feb 25Mar 4Mar 11Mar 18Mar 26
Cost Allocation
Chargeback by Business Unit
Cost Anomalies
▶ Show0 detectedOptimization Recommendations
▶ Show0 actions · $0 potential savingsAgent Cost Detail
36 Agents| Agent ▲ | Monthly Cost ▲ | Spend | Utilization ▲ | Cost/Invocation ▲ | Efficiency | Trend |
|---|
Access Control Matrix
Role-based permissions across all agents
Agent Operating Policies
Security policies defining what each agent can access, do, and escalate
14
Total AOP
10
Active
2
Draft
95.3%
Success Rate
Registered
37
Active agents
External
5
Platform integrations
Pending Review
2
Awaiting approval
Decommissioned
3
Last 90 days
Non-Compliant
1
Needs attention
➕ Register New Agent
🔗 Platform Integrations
Agent Lifecycle
0 agents| Agent | Category | System | Status | Runtime | Registered | Owner | Actions |
|---|
Amazon Bedrock AgentCore
Fetching...Runtime semiAgentRuntime-FLXK3hDb4x
Endpoint semiAgentEndpoint
Model Claude Sonnet 4
Region us-east-1
Agents 36 registered
—
Invocations (1h)
—
Avg Latency (ms)
—
Total Errors
—
Success Rate
28
Sessions
1
Throttles
0
System Errors
1
User Errors
0.056
CPU (vCPU-hrs)
0.211
Memory (GB-hrs)
7
Active WS Conns
113K / 404K
Bytes In/Out
Invocation Timeline (5-min)
Latency Distribution
⚡ Per-Agent Runtime Metrics
| Agent | Invocations | Avg Latency | Error Rate | Tool Calls | Tokens | Status |
|---|
🔍 Trace Explorer
38
Memory Ops (1h)
1,247
Sessions Stored
3,842
KB Queries
12.4GB
Vector Store Size
94.2%
Cache Hit Rate
847
Embeddings/hr
23ms
Avg Retrieval
4
Knowledge Bases
156K
Documents Indexed
0.89
Avg Relevance
2.1GB
Session Memory
3
Stale Entries
2,847
Requests (1h)
7
Active WS Conns
113K
Bytes In
404K
Bytes Out
99.8%
Uptime
42ms
P50 Latency
189ms
P99 Latency
1
Throttles
3
Endpoints
TLS 1.3
Protocol
312
Cross-Agent Calls
24
Unique Callers
108
Tool Invocations
14
Unique Tools
89
Code Interpreter
97.2%
Tool Success Rate
340ms
Avg Tool Latency
23
Lambda Functions
7
API Connectors
3,842
KB Retrievals
1.2s
Max Tool Time
2
Tool Timeouts
38.5K
Tokens (in)
16.0K
Tokens (out)
31
Registered Agents
28
Active Sessions
16
IAM Roles
4
Service Accounts
100%
Auth Success
0
Auth Failures
OAuth 2.0
Auth Protocol
mTLS
Agent-to-Agent
12
API Keys Active
3
Keys Expiring Soon
847
Token Refreshes
24h
Session TTL
1,847
Guardrail Checks
1,204
Content Filtered
423
PII Redacted
147
Denied Topics
73
Word Filters
99.1%
Compliance Rate
23
Shutdowns Prevented
8
Active Guardrails
3
Custom Policies
HIPAA
Compliance Std
SOC2
Audit Framework
0
Policy Violations
📋 Node Inspector
Click a node to inspect
⏱ Execution Log
Waiting for run…
Legend
⬤ Start / End
⬤ LLM Call
⬤ Tool
⬤ Guardrail
⬤ Condition
⬤ Error
Conversation Store
0—
Total Conversations
—
Avg Latency (ms)
—
Errors
—
Avg Tokens
Centralized Audit Log
0| Time | Action | Actor | Target | Details | Severity |
|---|
AgentCore Evaluations
0▶ Execute
📊 Results
🚦 Approval Gate 0
📜 Deploy History
▶ Execute Evaluation
🚦 Deployment Gate — Full Test Suite
—
Passed
—
Failed
—
Total Evals
—
Pass Rate
📋
No pending approvals. Run a full test suite and submit for review.Prompt Lab
Version Control & A/B Testing
vs
Version A
v3 · 2026-03-20
Version B
v2 · 2026-03-14
Diff:
Test Input
Output A
Run a test to see output…
Output B
Run a test to see output…
📊 Comparison Results
—
Latency A
—
Latency B
—
Tokens A
—
Tokens B
—
Cost A
—
Cost B
—
Relevance A
—
Relevance B
—
Safety A
—
Safety B
📜 Version History
Datasets & Experiments
Golden Test Suites🗂 Datasets
🧪 Experiments
📊 Compare
—
Datasets
—
Total Examples
—
Experiments Run
—
Last Run
Exp A:
Exp B:
Annotation Queue
Human-in-the-Loop Review—
In Queue
—
Approved
—
Rejected
—
Flagged
—
Agreement Rate
—
Reviewed Today
●
—
— · —
—
|
—
User Input
—
Agent Output
—
Latency: —
Tokens: —
Tools: —
Cost: —
Guardrail: —
Score Rubric
📋 Recent Reviews
Agent Chat
Powered by Amazon BedrockAsk me about your industrial-gas agents — status, metrics, compliance, or anything else.