Introducing Fovea Analytics: Privacy-First Analytics, Kubernetes-Native

We're building Fovea Analytics: privacy-first, cookieless web and product analytics that's Kubernetes-native by design. It's in active development. Here's what it is and why.

The Wedge

Analytics tools force a trade-off. Plausible is clean but ships as Docker Compose. PostHog is powerful but no longer supports new Kubernetes self-hosts and gates features behind Cloud.

We think there's an underserved middle: a CI-tested, signed Helm chart a K8s shop can helm install on any conformant cluster — EKS, GKE, AKS, k3s, or on-prem — for full data sovereignty. Our own SaaS runs on AKS, but that's a binding, not a constraint. The chart never assumes it.

Three commitments:

• Cloud-agnostic and self-hostable. Every component we run is the same portable component we ship. If we don't dogfood the chart, we're not selling what we run.
• Privacy-first and cookieless. No consent banner for the anonymous tier. EU residency by default.
• AI-native insights. Natural-language queries and anomaly detection — with the data model built for it from day one, not bolted on later.

The Stack

The architecture is deliberately boring where it counts and portable everywhere:

• Collector — Go on the hot path: validate, rate-limit, filter bots, hash visitors, drop IPs, sessionize, enqueue, and return.
• Event store — ClickHouse via the Altinity operator, with a schema pre-provisioned for identity and typed properties.
• Durable buffer — a Kafka-API log (Redpanda or NATS JetStream in the chart) as a crash-safety backstop.
• Sessions — Redis (HA). Control plane — Postgres. Backups — S3-compatible object storage.
• App and dashboard — Next.js, with a single tenant-scoped query chokepoint.

Multi-tenant from day one. Web analytics first, product analytics next — additive, not a rewrite.

What It's Not

Not for v1: feature flags, experiments, or session replay. Those are explicit non-goals, not missing features.

We'll share more as it ships. If you run Kubernetes and care about owning your analytics data, reach out — we'd like early feedback.