Installation Overview

Five sequential steps to install a self-hosted Prisme.ai platform, each linking to the next.

A self-hosted Prisme.ai install has five sequential steps. Each step links to the next, so you can follow the whole flow without jumping around the menu.

1. Plan

   Review the Requirements and Resources & Autoscaling to understand what to provision.

   → Next: provision your cloud environment.

2. Provision the cloud environment

   Before installing Prisme.ai with Helm, you need a Kubernetes cluster, managed databases, object storage and a load balancer in place. Provider-specific instructions:

   • AWS — EKS, RDS / Atlas, OpenSearch, ElastiCache, S3 (+ optional CloudFront), EFS, ALB.
   • Azure — AKS, CosmosDB / Postgres Flexible Server, Elastic Cloud, Azure Managed Redis, Blob, Azure Files ZRS, AGIC.
   • GCP — GKE, Cloud SQL / Atlas, Memorystore, GCS, Filestore regional, GKE Ingress.
   • OpenShift — OCP cluster, operators for Mongo / Redis / ES, ODF for object + file storage, Routes.

   Per-engine details (versions, sizing, least privileges): Databases.

   → Next: install Prisme.ai with Helm.

3. Install with Helm

   Install with Helm is the single source of truth for the Helm-level configuration. It covers both namespaces (core and apps), the critical values, secrets and ingress.

   The values you set here wire Prisme.ai to the cloud resources from step 2 (DB URLs, S3 buckets, ingress hostnames, etc.).

   Most of the tuning that follows — auth, mail, retention, resource requests, HPA targets, SSO, custom code — lives in the same prismeai-core-values.yaml / prismeai-apps-values.yaml and is applied by re-running helm upgrade -f <values>.

   → Next: tune the same Helm values and install products.

4. Tune the platform

   All four bullets below are additional Helm values layered onto the files from step 3 — edit, then helm upgrade -f to apply.

   • Configuration — the most important post-install settings (auth, mail, retention, upload limits, crawler, runtime).
   • Resources & Autoscaling — pick Balanced or High-throughput, HPA targets, per-service requests/limits.
   • Environment variables reference — full per-service env-var list, all settable under <service>.env in your values file.
   • Enterprise Services — SSO, RBAC, custom code, crawler.

   → Next: operate and install products.

5. Operate

   • Install products — provision the platform workspaces (Governe, LLM Gateway, Storage, Agent Creator, Insights, Builder, Helper Agents).
   • Migration to v27 if you’re upgrading from a legacy install.
   • Updates, Backup, Scaling.
   • Network Policies — minimal service, database and external egress flows for default-deny clusters (applied as separate Kubernetes resources, not via Helm values).