Representative sample · synthetic values
Cloud & AI economics sample model
A practical structure for linking spend, ownership, action risk and post-change verification.
Synthetic model · not customer evidence
Cloud-AI action register
The values below are illustrative. The model demonstrates how an action is linked to an owner, risk and verification rule without implying an achieved saving.
| Cost pool | Observed baseline | Candidate action | Owner | Verification |
|---|---|---|---|---|
| Model inference | ₹500,000/month (synthetic) | Route low-risk requests to a smaller evaluated model | Product + ML | Cost/request and quality gate |
| GPU training | ₹320,000/month (synthetic) | Schedule idle windows and require experiment owner | ML platform | Utilization and run ledger |
| Vector/data storage | ₹140,000/month (synthetic) | Apply retention tier after evidence review | Data owner | Storage trend and retrieval test |
Decision rule
No action is accepted from percentage savings alone. The register records expected benefit, delivery cost, service risk, data risk, dependency, owner and post-change evidence. Forecasts remain assumptions until the measured period closes.