Cast AI - EKS fully automated cost optimization and monitoring

What do you like best about the product?

The cluster looks healthy, but I’m still digging into what’s actually behind that green status. I find myself jumping between workloads, tracing resource behavior, and gradually the performance picture starts to make sense. I can take a quick look at the user interface, make a small adjustment, then go right back to the workload, and that rhythm genuinely suits the way I troubleshoot infrastructure.

What do you dislike about the product?

The onboarding feels too basic for an experienced engineer, with explanations that cover things I already know. On top of that, one integration behaves differently than I expected, which adds an unnecessary pause and interrupts the workflow.

What problems is the product solving and how is that benefiting you?

We were constantly chasing wasted capacity across our Kubernetes clusters, and our workloads needed manual tuning whenever traffic shifted. Cast AI started giving us AI-driven recommendations for resource allocation, and the ROI became clear once we applied those changes and began cutting unnecessary cloud spend.

What do you like best about the product?

I like CAST AI for its innovative approach to multi-cloud GPU capabilities, which allow me to choose GPUs across multiple cloud providers and even on-prem. This feature has been particularly beneficial for us. I also appreciate the resource optimizations, with memory right-sizing helping us identify workloads that need additional capacity. The automated workload optimization and extra insights on GPU optimization are also valuable. Additionally, the setup was very easy, and the customer support from CAST AI has been great.

What do you dislike about the product?

I think one area where CAST AI could improve is by increasing the transparency of its optimization decisions. It would be really helpful if the platform could offer richer insights into CPU and memory performance, especially around memory kills. This would allow us to proactively right-size the nodes. Having dashboards, alerts, and data-driven usage insights would also be beneficial. For example, understanding how minimum and maximum node settings impact applications that can have irregular operations throughout the month would be great.

What problems is the product solving and how is that benefiting you?

I use CAST AI for automated workload optimization and resource right-sizing. It helps identify workloads needing more capacity and offers insights on GPU optimization across multi-cloud providers.

What do you like best about the product?

One of the strengths of CAST AI is its cost optimization feature. I really like how it reduces costs using bin-packing technology and monitors available resources worldwide to operate at the most optimal cost.

What do you dislike about the product?

Due to different infrastructure configurations in each company, security settings were not easy. In particular, more integration capability is needed to resolve conflicts with security solutions. The initial setup was fine in the development network, but it took time to resolve firewall or security solution conflicts in the production network.

What problems is the product solving and how is that benefiting you?

The CAST AI in use maintains system stability while optimizing costs. It allows monitoring of available resources worldwide at the most efficient cost across various infrastructure configurations.

What do you like best about the product?

Its automated Kubernetes cost optimization which helped reduce cloud spend, improves resource efficiency, and removes much of the operational overhead of managing cluster scaling and infrastructure.

What do you dislike about the product?

The biggest downside is the learning curve, along with the need to build trust in the automation. Teams may need some time to understand how the optimization decisions are made and how to configure the platform in the best way for their environment.

What problems is the product solving and how is that benefiting you?

Cast AI addresses the challenge of Kubernetes cost optimization and infrastructure management. It helps identify overprovisioned resources, automate scaling, optimize node usage, and take advantage of Spot Instances to reduce cloud spend. As a result, we spend less time manually tuning clusters, get stronger resource utilization, and have clearer visibility into our cloud costs while still maintaining application performance.

What do you like best about the product?

Automatically cuts our Kubernetes costs without hurting speed.

What do you dislike about the product?

The UI is getting crowded. With security scanning, rightsizing, and spend forecasting added, finding a simple autoscaler setting can take three menus, and the docs seem to lag behind new features. I've had to contact support a few times, though they do respond. Node Templates also have a steep learning curve. You need to understand tolerations and node affinities to keep sensitive workloads like Redis and API gateways off spot nodes. Early on, a rebalance during a traffic spike briefly slowed our Retool dashboard. The savings-based pricing is fair to me as an engineer, but it's harder to explain to finance than a fixed license.

What problems is the product solving and how is that benefiting you?

Our biggest problem was a Kubernetes bill we couldn't explain. We run self-hosted inference, vector databases, and background jobs behind n8n and Retool, and I over-provisioned out of fear of latency spikes and cold starts. High-memory nodes and GPUs ran around the clock while sitting idle most of the time. Tuning requests and limits by hand was painful, and the native cluster autoscaler was too slow.
CAST AI fixed this in three ways. First, it showed us the waste. Our GPU utilization was in the single digits, and the projected savings were 40-50%. Second, it made spot instances usable. Batch document processing, transcription, and embedding jobs now run on cheap spot nodes, and it drains nodes gracefully before interruptions, so we don't get failed jobs or lost requests. Third, its bin packing keeps consolidating workloads onto the cheapest nodes, so we're not paying for idle capacity.
The result is lower compute costs, and I get my time back. I no longer babysit Grafana dashboards or explain why our GCP bill doubled after deploying a new model. Our real-time inference endpoints stay responsive too, as long as pod disruption budgets are set up properly.