Cast AI

Cast AI is an Application Performance Automation platform built by the team behind Zenedge (acquired by Oracle in 2018). It continuously monitors SLO signals such as error rates, latency, and OOM kills, then acts autonomously to fix issues before users are impacted. The platform is actively used and the company reports serving over 2,100 companies globally, according to its website.

What It Is

Cast AI sits in the Kubernetes cost optimization and infrastructure automation category. Its core job is to replace manual cloud operations with an autonomous engine that scales, rightsizes, and rebalances workloads in real time. Rather than surfacing recommendations for engineers to act on, Cast AI executes changes directly — adjusting CPU and memory requests at the millicore level, managing spot instance lifecycles, and placing pods on optimal instance types. The platform covers the full stack from infrastructure (node scaling, GPU allocation) to application-layer concerns (database optimization, LLM cost optimization for AIOps).

How the Automation Engine Works

The Cast AI Engine is built on a predictive model trained on data from thousands of clusters and millions of real-world workloads. The platform describes four operational modes:

• Connect — Deploy to Kubernetes clusters in minutes, starting in read-only mode with no infrastructure changes required.
• Analyze — Observe real workload behavior (not static configurations) to identify optimization opportunities.
• Optimize — Automatically scale, rightsize, and rebalance based on real-time signals rather than scheduled jobs.
• Fix — Use agentic runbooks to remediate operational and security issues; users approve changes before they ship.

The engine claims to predict spot interruptions up to 30 minutes before they occur, migrating workloads gracefully before users experience slowdowns.

Platform Capabilities

Cast AI organizes its feature set into four pillars:

• Self-healing operations — AI agents that remediate drift, update container images, auto-heal failures, and enforce policies without manual tickets.
• Performance observability & intelligence — Real-time visibility into resource utilization and application performance.
• Workload rightsizing — Automatic adjustment of CPU and memory requests to match actual usage, eliminating over-provisioning.
• Infrastructure automation — Node scaling, GPU allocation (OMNI Compute), spot instance management, and a single control plane across any cloud or on-premises environment.

The platform also includes dedicated modules for Karpenter optimization, database optimization, and LLM cost optimization for AIOps workloads.

Cloud and Tool Integrations

Cast AI integrates with all three major cloud providers — AWS (EKS), Google Cloud (GKE), and Azure (AKS) — as well as Oracle Cloud and IBM Cloud. On the tooling side, it connects with Kubernetes, Grafana, Terraform, Crossplane, Prometheus, OpenTelemetry, Helm, Rancher, OpenShift, VMware Tanzu, kOps, AliCloud, Pulumi, Jira, PostgreSQL, and MySQL, among others. The company holds ISO 27001 and SOC 2 certifications and is a member of the CNCF and FinOps Foundation.

Founding Story and Team

Cast AI was founded by Yuri Frayman (CEO), Leon Kuperman (CTO), and Laurent Gil (President) after their previous company Zenedge was acquired by Oracle in 2018. The founders describe the product as originating from personal frustration with growing cloud bills after that acquisition. The company has offices in Miami, New York, Vilnius, Tel Aviv, London, Dallas, and Bengaluru, and is backed by investors including Creandum, Samsung Next, SoftBank, and Hedosophia.