Datasets:
roanbrasil
/
devops-gitops-corpus

Dataset card Data Studio Files
xet
 Community
1
text
stringlengths
0
59.1k
graph TB
A[User Input] --> B[Agent Processing]
subgraph "Observability Layer"
B --> C[Context Analysis]
C --> D[Decision Making]
D --> E[Tool Selection]
E --> F[External Execution]
F --> G[Response Generation]
end
G --> H[User Response]
C --> I[Memory Logs]
D --> J[Reasoning Logs]
E --> K[Decision Logs]
F --> L[Tool Logs]
G --> M[Output Logs]
classDef visible fill:#10b981,color:#ffffff,font-size:10px
classDef logs fill:#34d399,color:#064e3b,font-size:10px
classDef input fill:#d1fae5,color:#064e3b,font-size:10px
class C,D,E,F,G visible
class I,J,K,L,M logs
class A,B,H input
`} />
This visibility is critical to:
- **Fixing deep, multi-component bugs**
- **Understanding performance** bottlenecks
- **Optimizing behavior** based on observed patterns
- **Building confidence** in AI decisions
## Real-World Example: Debugging with Observability
Here's a walk-through of how observability tools can help solve real-world testing problems.
### Example scenario: Multi-step AI workflow
Let's say we're analyzing a customer support AI that can:
- Search customer information
- Check order status
- Schedule follow-up calls
- Send email notifications
### Observability-driven debugging process
When analyzing the query: "Hi, I'm john.doe@email.com and need help with my previous order"
Modern observability shows the complete execution flow:
#### Step 1: Initial Processing
```json
{
"step": 1,
"type": "user_input",
"content": "Hi, I'm john.doe@email.com and need help with my previous order",
"timestamp": "2024-01-15T10:00:00Z",
"extracted_entities": ["email", "order_inquiry"]
}
```
#### Step 2: Decision Analysis
```json
{
"step": 2,
"type": "decision_making",
"decision": "customer_lookup_required",
"confidence": 0.95,
"reasoning": "Email provided, need profile before proceeding with order inquiry"
}
```
#### Step 3: Customer Lookup
```json
{
"step": 3,
"type": "external_call",
"service": "customer_database",
"input": {
"email": "john.doe@email.com"
},
"output": {
"customerId": "cust_123",
"name": "John Doe",
"tier": "premium",
"lastContact": "2024-01-10"
},
"executionTime": "156ms"
}
```
### Common problems revealed through observability