.png)
CAST AI focuses on Kubernetes optimization and autoscaling, while Usage.ai focuses on cloud commitment management.
That distinction matters when planning a migration. Your existing commitments, Kubernetes automation, and purchasing workflows may all be connected, so removing one tool before understanding those dependencies can create unnecessary risk.
In this guide, we’ll walk through what CAST AI is managing today, how to evaluate Usage.ai for commitment management, how to compare the economics, and how to make the handoff without disrupting your Kubernetes environment.
You can also keep CAST AI for Kubernetes optimization while moving commitment management to Usage.ai.
If your goal is to leave CAST AI completely, we’ll treat that as a separate Kubernetes migration rather than mixing the two changes together.
The safest approach is:
Migration at a glance
| Phase | What you do | Output |
|---|---|---|
| Baseline | Record cloud spend, commitments, utilization, and CAST AI savings | Comparable starting point |
| Dependency review | Map how CAST AI uses commitments and manages Kubernetes | Clear migration scope |
| Portfolio review | Classify existing commitments | Keep/reassess list |
| Usage.ai evaluation | Evaluate commitment opportunities before purchasing | New opportunity baseline |
| Economics | Compare current, Usage.ai, and combined scenarios | Go/no-go decision |
| Handoff | Establish one purchasing authority | Controlled cutover |
| Verification | Monitor billing, utilization, and savings | Validated migration |
| Offboarding | Remove CAST AI only if appropriate | Completed transition |
Before you migrate: establish your baseline
Don’t disconnect CAST AI and then try to reconstruct what it was doing.Start by recording the current state of both your cloud economics and Kubernetes environment.
Capture your cloud baseline
Cloud spend for a consistent recent period
RI, Savings Plan, or other applicable commitment inventory
Coverage and utilization
Remaining terms and expiration dates
Services and accounts included
Current commitment costs
Capture your CAST AI baseline
Record:
CAST AI-reported savings
Node optimization impact
Workload optimization impact
Spot-related savings
Commitment-related savings or utilization
CAST AI fees
Capture your Kubernetes dependencies
Document the CAST AI functions currently enabled in production, including:Node autoscaling
Workload optimization
Spot management
Node provisioning
Commitment-related scaling behavior
Terraform ownership, where applicable
The goal is to know what you’re getting from CAST AI before you decide what you’re willing to give up.
Understand how CAST AI currently uses commitments
This is the most important CAST AI-specific step.According to CAST AI’s Commitments documentation, CAST AI can import supported cloud commitments, assign them to clusters, track their utilization, and use them when making autoscaling decisions. For AWS, this includes Reserved Instances and Savings Plans.
Before changing anything, find out exactly how those commitments are being used today.
Ask:
Which commitments are being used by CAST AI-managed workloads?
Do any of those commitments also cover workloads outside Kubernetes?
Which commitments are healthy, and which are already underutilized?
Who will manage commitment purchases after the migration?
Who will continue making Kubernetes capacity and scaling decisions?
CAST AI also supports commitment utilization across workloads outside CAST AI-managed clusters. So don’t assume that a commitment used by a CAST AI cluster is exclusively tied to that cluster.
The key is to separate two decisions:
- Commitment management: how much coverage to buy, when to buy it, and how to manage the financial risk.
- Kubernetes optimization: how much capacity to run and how workloads should be scaled.
Classify your existing commitments
Don’t treat your existing portfolio as something you need to replace.Separate it into three groups.
Healthy commitments
Well-utilized commitments with meaningful time remaining can generally continue operating. Changing the management process doesn’t automatically make an existing commitment uneconomic.Commitments nearing expiration
These are natural reassessment points. Compare the economics of renewing them under your current approach with the opportunities identified through Usage.ai.Underutilized commitments
These require closer review. Changing your optimization platform does not automatically remove the underlying cloud-provider commitment.The objective is to distinguish existing exposure from future purchasing decisions.
Evaluate Usage.ai before changing purchasing authority
Once the baseline is established, evaluate what Usage.ai could add.At Usage.ai, we recommend starting with the evaluation rather than immediately changing purchasing permissions.
Our Security and Compliance documentation describes the read-only permissions available for Savings Tests. With read-only access enabled, Usage.ai can evaluate potential savings without the ability to purchase Reserved Instances or Savings Plans.
Use your current cloud-spend and commitment baseline to assess:
Coverage: Where are the current gaps?
Existing commitments: Which commitments are already performing well?
New opportunities: Which eligible commitments could improve the economics?
Usage assumptions: What usage supports each recommendation?
Net savings: What remains after applicable Usage.ai fees?
Downside: What happens if committed usage falls?
Protection: Which eligible Flex Commitments qualify for applicable Cashback Protection?
The purpose isn’t to prove that Usage.ai will always save more.
It’s to determine whether Usage.ai creates enough incremental value to justify changing your operating model.
Also read: Usage.ai vs CAST AI: Which Fits Your Savings Strategy?
Compare the economics, not the savings percentages
Once you know what CAST AI is managing today, compare the net economics of each option.Don’t compare one vendor’s headline savings percentage with another’s. Instead, build at least three scenarios:
| Scenario | What you're measuring |
|---|---|
| CAST AI today | Kubernetes optimization + current commitment economics − applicable CAST AI costs |
| Usage.ai | Commitment savings − applicable Usage.ai fees + applicable protection |
| CAST AI + Usage.ai | CAST AI Kubernetes optimization + Usage.ai commitment economics − applicable costs |
If CAST AI is delivering meaningful value through Kubernetes optimization, you may not want to replace it. You may be able to keep that optimization in place while moving commitment management to Usage.ai.
A simple example
Suppose you have $100,000 in monthly eligible cloud spend and CAST AI is currently delivering $15,000 in total optimization value.Don’t treat that $15,000 as a single savings number to beat.
First, separate the value coming from Kubernetes optimization from the value coming from commitment management. Then ask:
What value do you retain by staying with CAST AI?
What additional commitment savings could Usage.ai provide?
What does the combined CAST AI + Usage.ai model look like?
What fees and other costs apply to each option?
What happens if your usage changes?
| Scenario | Question |
|---|---|
| Base | What if usage stays broadly consistent? |
| Downside | What if committed usage falls materially? |
| Growth | What if usage grows faster than expected? |
How to migrate from CAST AI to Usage.ai
Once the economics support the move, keep the operational handoff controlled.1. Lock the baseline
Save your cloud-spend, commitment, utilization, and CAST AI savings data.Make sure Finance, FinOps, and Engineering agree on the baseline you’re going to use for measuring the migration.
2. Map CAST AI dependencies
Identify which Kubernetes functions remain with CAST AI and which, if any, you intend to replace.If CAST AI remains, document how its existing commitment utilization will interact with the new purchasing process.
3. Review the existing commitment portfolio
Identify healthy, expiring, and underutilized commitments.Don’t assume changing management means replacing the existing portfolio.
4. Run the Usage.ai evaluation
Evaluate the available commitment opportunities against your current usage and existing coverage.Start with the least-privileged access appropriate to your evaluation and review the recommendations before granting purchasing authority.
5. Validate the economics
Compare the current CAST AI model, Usage.ai model, and combined model against the same baseline.If the combined approach produces the strongest net result, there may be no economic reason to remove CAST AI.
6. Establish one purchasing authority
Before enabling a new purchasing workflow, clearly define which system or team is authorized to make new commitment purchases.Avoid a period where two systems can independently make commitment decisions against the same portfolio.
7. Make the cutover
Approve the new commitment strategy through your normal FinOps, Finance, or procurement process.Verify new purchases directly with the cloud provider.
If CAST AI remains active, confirm that its existing Kubernetes optimization continues to behave as expected.
8. Verify the results
Don’t call the migration complete simply because the integration is enabled.Monitor:
Commitment coverage
Commitment utilization
Cloud billing
Realized savings
Usage.ai fees
Applicable cashback
CAST AI performance, if it remains in place
What if you want to fully leave CAST AI?
That’s a separate migration.Before disconnecting CAST AI, identify replacements for the Kubernetes capabilities it currently manages, including node autoscaling, workload optimization, Spot management, and node provisioning.
Do not combine an unplanned Kubernetes automation change with a commitment-management cutover.
CAST AI’s cluster disconnect documentation provides the current procedures. CAST AI notes that castctl cluster disconnect is not intended for Terraform-managed clusters; those environments should use the appropriate Terraform workflow.
Follow CAST AI’s current offboarding documentation for the environment you’re actually running.
A move to Usage.ai isn’t automatically the right decision.
Stay with CAST AI if Kubernetes optimization is producing most of your savings and its automation is delivering material value.
Evaluate Usage.ai if commitment economics are a significant opportunity or you want to separate commitment management from Kubernetes optimization.
Consider both if CAST AI is valuable for Kubernetes while commitment management represents a separate savings opportunity.
Which operating model gives us the best net economics without giving up optimization we already depend on?
CAST AI → Usage.ai migration checklist
Use this checklist to keep the migration controlled, measurable, and easy to validate at each stage.Establish a shared cloud-spend baseline
Preserve the CAST AI savings baseline
Inventory existing commitments
Map CAST AI's Kubernetes dependencies
Document current purchasing authority
Evaluate
Run the Usage.ai evaluation
Reconcile recommendations against existing commitments
Compare current, Usage.ai, and combined scenarios
Model base, downside, and growth cases
Confirm applicable program terms
Cut over
Establish one purchasing authority
Approve the new commitment strategy
Verify purchases with the cloud provider
Monitor billing, utilization, and savings
Review results against the baseline
Offboard only if appropriate
Replace required Kubernetes functions
Follow the applicable CAST AI disconnect process
Remove unnecessary permissions
Revalidate the environment
Should you migrate from CAST AI?
Not automatically.If CAST AI is delivering strong Kubernetes optimization and your main cost challenge isn’t commitment management, there may be little reason to make a disruptive change.
If commitment economics are a meaningful opportunity, however, Usage.ai gives you a way to evaluate that layer separately before deciding whether CAST AI itself needs to go.
And if both platforms can contribute distinct value, keeping CAST AI for Kubernetes optimization while using Usage.ai for commitment management may be the better outcome.
The best migration is the one your own cloud data supports.
Bring your current commitment inventory, recent cloud-spend data, CAST AI savings baseline, and one realistic downside scenario.
At Usage.ai, we’ll map your existing commitment coverage to your actual cloud spend, model the economics under base and downside scenarios, and show you where Usage.ai could
Talk to an Expert
Frequently asked questions
Can I use Usage.ai without removing CAST AI?
Yes. You can keep CAST AI for Kubernetes optimization and evaluate Usage.ai separately for cloud commitment management. This can be a lower-risk approach if CAST AI is already delivering meaningful Kubernetes savings.
Do I need to replace my existing cloud commitments?
Not necessarily. First inventory your existing commitments and their utilization. Avoid replacing healthy commitments simply because you're changing the system that manages future commitment decisions.
Will migrating to Usage.ai affect my Kubernetes workloads?
A commitment-management migration does not require you to migrate your Kubernetes workloads. If you decide to fully leave CAST AI, treat Kubernetes optimization and cluster offboarding as a separate migration.
How should I compare CAST AI and Usage.ai?
Compare net economics, not headline savings percentages. Model what you save today, what Usage.ai could add through commitment optimization, applicable fees, and the outcome under base, downside, and growth scenarios.
How do I know when it's safe to make the switch?
Set a baseline, validate your existing commitments, run the Usage.ai evaluation, and agree on who owns future commitment purchases before the cutover. After the handoff, verify coverage, utilization, and realized economics before making broader changes.
