Enterprise AI platform
Cost savings
Time to market
Security
Not Just a Test Playground, but an All-in-One Enterprise AI Platform In Your Environment
How it Works?
Our Technology Stack Components
Manager
Admin backend for the developer portal and orchestrating deployments.
- Handles creating pipeline-runs, authorization policies, user access roles, etc.
- All writes to the master DB (both user and application writes) are handled by this service.
Realtime
Listen/notify implementation for the master DB.
The developer portal and deployed services subscribe to updates for the services they have access to.
Admin portal
Admin user interface for managing the stack.
Non-admin users see an "AppStore" like grid-view of the end-user applications they have access to.
CI/CD
- Tekton core stack in the "tekton-pipelines" namespace.
- Allows building and deploying services from source code or image registry. It is used for all of our internal services and images, and for deploying services from the manager.
- If you make custom changes to pipelines or tasks, you have to run
helm upgrade stack-basebefore changes take effect. - A separate helm chart is used to define the infrastructure needed for services deployed by the CI/CD.
Traffic management and authz
Istio provides a service mesh for internal and external traffic management, security, and observability.
- For external communication, it's used for JWT token checks, mapping traffic to correct services, and handling TLS certs together with cert-manager.
- For internal communication, Istio provides mutual TLS and service discovery inside the stack.
- "Istio-system" namespace for Istio base resources deployment.
- "istio-ingress" namespace for gateway deployment.
- All namespaces with Istio-sidecar enabled will have Istio Envoy sidecar automatically deployed to enforce mTLS.
User management & authentication
- Keycloak provides a full user management and authentication service.
- It uses CNPG Postgre cluster as its database.
- Initialized with realm-export.json to create correct realms and clients on install.
Databases
- Stack uses CNPG operator that manages the deployment of Postgre clusters.
- Both the master database and Keycloak database are deployed as part of the stack-base helm chart using the operator.
- Capability to deploy separate Postgre clusters for applications with pg-vector and other extensions.
Cert Management
Allows automatic cert management using LetsEncrypt.
- For this, the stack needs a public IP and correct DNS setup.
Providing certificates manually is also possible.
An enterprise AI platform is a collection of microservices, security protocols, and development tools to enable AI capabilities for organizations and nation states. With the recent emergence of generative AI, these types of platforms have become essential for fast prototyping and production without spending years developing underlying software infrastructure.
When choosing an enterprise AI platform, here are the things you need:
- Enterprise microservices
- Simple UI
- Cloud-agnostic
- Edge computing capabilities
- Data security and governance
- Unified data aggregation
- Authorization and authentication
- Effortless development for software developers
- On-premises and private cloud environments
We live in a world where AI forces enterprises to reimagine their workflows how they work. As a result, AI will be integrated into each system, department, and workflow. The key to this new digital disruption is the speed of adoption. Because the worst-case scenario is a slow reaction which can result in a loss of competitive advantage.
Enterprise AI platforms offer a solution for the complexity in AI adoption. These platforms help to increase the speed of adoption, because building technology infrastructure from the ground up is complicated and requires many specialized roles. The speed of adoption of this new digital disruption makes or breaks organizations.
ConfidentialMind's enterprise AI platform is designed with security in mind. We focus on the most demanding data security scenarios in on-premises and private cloud environments. That said, we have simplified and abstracted the complexity so that every software developer can start the process. Say goodbye to years of stress and complex infrastructure development and hello to quickly developing your PoCs and effortlessly move to enterprise-wide production once ready.
