Akamas

Akamas is an AI-powered platform built by Akamas S.p.A. (Milan, Italy) that delivers continuous, autonomous optimization across the full cloud-native stack — from Kubernetes pod resources and HPA scaling settings to JVM/Node.js runtimes and cloud instance selection. The platform uses patented reinforcement learning algorithms to analyze real workload behavior, identify inefficiencies, and apply or recommend safe configuration changes without requiring manual tuning cycles.

What It Is

Akamas sits at the intersection of performance engineering and platform automation. Rather than offering static rules or one-time recommendations, it continuously learns from live telemetry and iteratively explores configuration options to find settings that simultaneously improve performance, reliability, and cost efficiency. The platform offers two primary modules: Insights for production optimization with real-time recommendations, and Offline for automated pre-production performance testing and tuning studies.

How the Full-Stack Optimization Works

Akamas connects to existing observability tooling (Dynatrace, Datadog, Prometheus) and ingests telemetry to analyze workloads across multiple layers simultaneously:

• Runtime layer — JVM, Node.js, Go configuration parameters
• Application layer — microservices and batch job settings
• Kubernetes layer — pod resource requests/limits, HPA and KEDA scaling parameters
• Infrastructure layer — node group sizing and cloud instance selection

The platform's application-awareness means it considers cross-layer configuration interdependencies, avoiding changes that improve one metric at the expense of another. The About page notes the platform is "grounded in decades of applied performance engineering" originating from Moviri, where the core technology was developed.

Patented AI Engine

At the core of Akamas is a proprietary reinforcement learning engine. According to the company, this makes Akamas "one of the very few optimization platforms built on proprietary, peer-reviewed algorithms." The AI autonomously explores configuration spaces that have grown from roughly 20 parameters in 2015 to over 1,000 today in modern cloud-native environments, learning system behavior over time to avoid dangerous configuration changes in production.

Target Audience and Team Alignment

Akamas is designed to bridge the gap between developers, SREs, and platform engineers — teams that historically optimize in silos. The Insights module is positioned as a shared interface that aligns these roles around a common optimization goal. The platform also addresses what the company describes as a "cross-team skills gap" between application developers and Kubernetes administrators, surfacing actionable recommendations that either team can act on.

Deployment Model and Integrations

Akamas is delivered as a SaaS platform accessed via the web. It integrates with major observability platforms including Dynatrace, Datadog, and Prometheus, and supports GitOps workflows for deploying optimized configurations. The Offline module integrates with CI/CD pipelines to automate performance test execution and configuration scoring in pre-production environments.

Update: GigaOm Recognition and Recent Product Direction (2026)

The company's blog and newsroom show active product development into 2026. Recent releases highlighted on the site include HPA-aware optimization in Akamas Insights, Node.js full-stack optimization support, and Kubernetes tuning profiles. The company also published "The State of Cloud Native Optimization 2026" industry report and announced a June 2026 webinar on GenAI workload optimization on Kubernetes. GigaOm's 2026 Radar for Cloud Resource Optimization named Akamas both a Leader and Outperformer, a claim attributed to GigaOm on the Akamas website.