Instead of teams spending hours right-sizing nodes, tweaking autoscalers, or guessing instance types, Cast AI continuously analyzes workload requirements and automatically:

Selects the most cost-efficient instance types

Optimizes cluster autoscaling in real time

Uses spot instances intelligently to reduce costs

Eliminates overprovisioning without impacting performance Review collected by and hosted on G2.com.

Vendor dependency: Relying on automation to make core infrastructure decisions increases our operational dependence on a third party. Review collected by and hosted on G2.com.

Cast AI has delivered real, automated cost reductions in GCP for our highly scalable communications API without compromising reliability or control. It continuously optimizes our Kubernetes workloads in a way that is safe, predictable, and well-aligned with production requirements.

What I value most is that our SRE team stays firmly in control. Cast AI provides clear recommendations, enforces guardrails we define, and automates execution without becoming a black box.

We’ve seen meaningful infrastructure savings with zero production impact and minimal ongoing effort. For a team operating at scale, that balance of automation, safety, and control is the biggest win. Review collected by and hosted on G2.com.

The main gap is around higher-level visibility and reporting. While the optimization itself works well, turning the data into clear, executive-ready views of savings trends and attribution takes some extra effort.

Additionally, more built-in explanation behind certain recommendations would help with onboarding and broader team understanding. Improving reporting and explainability would make an already strong platform even better. Review collected by and hosted on G2.com.

I appreciate the range of diverse resources that the tool offers and, in addition to the constant updates focused on improving the tool, it is possible to notice the appreciation of those who work on this project and their commitment to excellence. The ability to identify how much each team's application running on the cluster costs is sensational. Review collected by and hosted on G2.com.

Sometimes, its scaling is not that interesting. I don't know if this happens because of the calculations of Requests and Limits, but it often tries to spin up some nodes with absurd amounts of memory, which would make me spend much more than save. Therefore, it is necessary to impose several limits and not let it run loose; otherwise, the cost may end up increasing instead of decreasing. Review collected by and hosted on G2.com.

I love how CAST AI doesn't just point out where I'm overspending, but actually goes in and automates the fixes so I don't have to manually tweak clusters all day. Its ability to handle Spot Instances is a total lifesaver because it keeps my apps running smoothly while cutting my cloud bill by more than half. Honestly, it’s like having a dedicated DevOps pro working 24/7 to make sure we’re never paying for more than we actually use. Review collected by and hosted on G2.com.

One thing I find frustrating is that giving up that much control to an autonomous "black box" can be nerve-wracking, especially since it occasionally restarts workloads to move them and the reporting isn't always as granular as the native cloud billing I'm used to. Review collected by and hosted on G2.com.

I appreciate the range of diverse features that CAST AI offers, and in addition to the constant updates focused on improving the tool, it's possible to perceive the appreciation of those working on this project and their commitment to excellence. I like how the tool revolves around AI, but it allows for fine-tuning of configurations to achieve excellence. The ability to allocate different machines to nodes within the same pool without requiring multiple configurations is particularly beneficial. The initial setup is super simple, basically running two or three commands on the cluster and all the configuration is ready. Review collected by and hosted on G2.com.

I believe the notification aspect of CAST AI could be improved; it's very simple, which is good for usability, but bad because it makes it limited. It would be interesting to have the possibility of receiving error alerts and alerts when recovering from those errors, allowing us to actively monitor the tool's operation. Review collected by and hosted on G2.com.

I really like the way CAST AI is designed because it aligns well with our company culture. Their support is also very helpful, which I appreciate. I find the AI support integration features beneficial for addressing simple questions and debugging scenarios. The ease of enabling features like vertical auto scaling and scheduling rebalancing is quite user-friendly. I also value the early access feature that allows us to hibernate our development and staging clusters during off hours, which is very helpful for us. And one big feature that helped us a lot was the ability to allocate a part of workload in to spot machines. Review collected by and hosted on G2.com.

There's definitely a learning curve to the platform, especially understanding the concepts and how it scales costs for you. Also, the billing aspect takes some time to understand. And some guarantees on the billing aspects (like if the cost reduction doesn't happen as expected) would reduce the initial inhibition of customers to adapt the system. For us personally another challenge was understanding how to measure cast AI's performance once we start using their system but also our own system starts growing. Actually there are indicators in the cast AI for this (like the cluster score, CPU & Memory over-provisioning percentage etc). Tracking these actually helps us understand if the cast AI's optimisations are as good as when we started. This could be documented as an indicator for cast AI's efficiency which will help the customers a lot. Review collected by and hosted on G2.com.

I really like the simplicity of setting up CAST AI; it was super easy and took less than an hour. We just followed the docs, and within minutes, everything was running smoothly. One of the standout benefits for us has been the cost reduction. We've managed to reduce our costs by a huge 60% since using it. I appreciate how CAST AI ensures that pod limits and requests align with real usage thanks to rightsizing. Previously, managing node pools manually was a hassle. With CAST AI, it's fully automatic, so we don't have to worry about manually creating new node pools to match our resource needs anymore. Review collected by and hosted on G2.com.

We had a small glitch at some point where the cluster scaled up because of a poorly configured workload but it didn't scale down automatically which end-up augmenting our costs a lot. We definitely need a way to have alerting when the daily price is more than x% of the previous days. Review collected by and hosted on G2.com.

I like the valuable cost insights dashboard that helps us track and analyze our cluster expenses over time along with future cost predictions, making it easier to set and maintain our cost goals. The node recommendations by CAST AI ensure workloads run efficiently. If a node is over-provisioned, it suggests the right sized instances. I enjoy the integration with multiple clouds, including AWS, Azure, and GCP, as the onboarding experience with CAST AI has been extremely smooth. Additionally, I appreciate that many teams in our organization have switched to CAST AI because it is efficient and saves time for our team. Review collected by and hosted on G2.com.

Sometimes the suggestions are too aggressive for nodes and it may lead to workload discrepancy. Review collected by and hosted on G2.com.

I use CAST AI primarily for node management and workload optimization within our Kubernetes clusters. I like its ability to automatically select the most cost-effective nodes for our workloads, taking a lot of guesswork out of infrastructure management. The intelligent instance selection feature dynamically chooses the optimal mix based on performance requirements and cost efficiency. Additionally, I appreciate its recommendations for right-sizing workloads, which help to save resources. The setup is also quick and easy, and onboarding clusters onto CAST AI is straightforward, with Terraform assisting in managing CAST AI resources. Review collected by and hosted on G2.com.

One area that could be improved is dynamic workload right-sizing. While it's useful, it heavily relies on past actual usage to forecast and adjust resource requests and limits. It works, but in cases where traffic spikes unusually, it doesn't always adapt quickly enough. Review collected by and hosted on G2.com.

I like the UI and the ease with which I can operate an EKS cluster on CAST AI. For example, rebalancing one of our largest clusters back to an optimized state took just a few clicks. Even someone who's not familiar with coding and Terraform can easily accomplish this task. I also like their support, particularly how I can interact with them via Slack. They respond extremely quickly, dive deep into any issues, and are available to jump on a call and explain things thoroughly. Additionally, Cast AI's solution comes with their own autoscaler, which allowed us to start seeing savings immediately. Their support was very helpful during the setup process, assisting us in all aspects. Review collected by and hosted on G2.com.

I think if there was an easier way to upgrade EKS clusters that would be a plus because currently after upgrading you still need to manually rebalance the nodes for all of them to get upgraded, where with EKS auto mode it's completely automated. Review collected by and hosted on G2.com.