AI System Map
System Context
- System:
- Primary user:
- Task or decision supported:
- Expected outcome:
- Unacceptable outcome:
- Accountable owner:
Request Flow
Map the path from user request to response or action. Include the interface,
application logic, context, model, tools, validation, and fallback that matter
for this system.
flowchart LR
A[User] --> B[Interface]
B --> C[Application]
C --> D[Context and retrieval]
D --> E[Model]
C --> F[Tools and business systems]
E --> G[Validation and fallback]
F --> G
G --> B
C --> H[Logs and evaluation]
Data and Knowledge Flow
| Data or content |
Source and owner |
Processing or storage |
Used when |
Freshness requirement |
| |
|
|
|
|
Components and Responsibilities
| Component |
Responsibility |
Input |
Output |
Owner |
Known or assumed? |
| |
|
|
|
|
|
Trust Boundaries and Permissions
| Boundary |
Data or authority crossing it |
Required control |
Failure impact |
| |
|
|
|
Human Oversight and Recovery
- When must a person review or approve?
- How can the user challenge or correct an output?
- What fallback exists when AI or a dependency fails?
- Which actions can be reversed?
Architecture Unknowns
| Unknown |
Why it matters |
Best source or owner to clarify it |
| |
|
|