- SaaS Managed by Prisme.ai: Shared or dedicated managed service hosted by Prisme.ai.
- Self-Hosted or Co-Managed: Deploy on your cloud of choice including AWS, GCP, Azure, OVH Cloud, or any cloud offering managed Kubernetes.
Why Choose Self-Hosting?
Choosing self-hosting provides your organization with:- Data Sovereignty: Keep your sensitive data within your control.
- Security Compliance: Meet industry-specific security and compliance requirements.
- Customization and Integration: Fully customize and integrate Prisme.ai into your existing workflows and systems.
- Cost Efficiency: Optimize resource allocation and management.
Deployment Options
Prisme.ai supports various deployment scenarios to suit your organization’s infrastructure strategy:- Cloud-Agnostic Deployment: Compatible with AWS, Azure, Google Cloud, and private cloud providers.
- Kubernetes-native: Deployed via Helm charts on any conformant Kubernetes cluster.
- High Availability: Designed for redundancy and resilience.
- Scalability: Easily scale horizontally and vertically based on workload.
Supported Environments
Prisme.ai self-hosting runs on any Kubernetes 1.26+ cluster, including:- Cloud provider-managed services like EKS, AKS, GKE, OVHcloud Managed Kubernetes.
- OpenShift.
- Self-managed Kubernetes on bare metal or VMs.
Platform microservices
Prisme.ai is deployed as two Helm releases, each in its own Kubernetes namespace.core namespace
Always required. Handles authentication, workspace storage, automation execution and the user-facing console.
| Service | Role |
|---|---|
prismeai-api-gateway | Single public entry point. Routes API calls, terminates authentication (local + SSO), implement all organization APIs, enforces org RBAC. |
prismeai-console | Frontend that serves the Studio (workspace builder) and the user-facing pages of each workspace. |
prismeai-workspaces | Authoritative store for workspace models (DSUL): create / read / update / version / publish workspaces and their automations, pages, blocks, custom domains. |
prismeai-runtime | Automation execution engine. Runs the event-driven workflows of every workspace, manages execution contexts, evaluates conditions and templates. Multi-threaded. |
prismeai-events | Real-time event bus over Redis Streams + persistence in Elasticsearch or OpenSearch. Powers the activity feed, analytics and audit. Hosts the cleanup CronJobs. |
apps namespace
Optional services that extend the platform with custom code, web crawling and search.
| Service | Role |
|---|---|
prismeai-functions | Sandboxed execution of user-defined JavaScript and Python code (custom automations, AI Products business logic). Ships pre-installed JS / Python dependencies — see Custom Code Functions. |
prismeai-crawler | Web and document crawler. Fetches, parses and indexes content into Elasticsearch or OpenSearch for RAG use cases. |
prismeai-searchengine | Query layer on top of crawler indexes. Exposes a stable search API and ranking to the products and to runtime automations. |
Next Steps
Installation Overview
Five-step install flow, from cloud provisioning to operations.
Requirements
Infrastructure prerequisites, sizing baseline, networking.
Install with Helm
Unified Helm install reference for core and apps namespaces.
Databases
PostgreSQL or MongoDB, Redis, Elasticsearch or OpenSearch.
Need detailed architecture diagrams (draw.io) tailored to your deployment? Contact support@prisme.ai.