CrewAI Factory
Product · Self-hosted / on-prem software by CrewAI
The customer-managed AI factory for enterprise-scale agentic workflows.
CrewAI Factory is a customer-managed deployment of the CrewAI Enterprise platform for organizations requiring on-premises or VPC-based AI agent orchestration. It integrates NVIDIA AI Enterprise to provide specialized models and high-performance inference on private infrastructure.
- Launched
- 2025-05-19
- Languages
- English
Category: AI Infrastructure
- Compatibility
- NVIDIA AI Enterprise, NVIDIA NeMo, NVIDIA NIM microservices
- License
- Proprietary (Enterprise license)
- Supported Cloud Providers
- AWS, Azure, Google Cloud
- Container Services
- AWS Fargate, Azure Container Apps
- Orchestration Tools
- Kubernetes, Helm, Red Hat OpenShift
- Database Compatibility
- AWS RDS, Azure Database
- Customer-managed deployment on private data centers or VPCs
- Integrated with NVIDIA AI Enterprise and NIM microservices
- Support for Kubernetes, AWS Fargate, and Azure Container Apps
- Native integration with Auth0 and Microsoft Entra ID
Overview
It allows businesses to deploy the full Agent Management Platform within their own data centers or virtual private clouds (VPC), supporting environments like AWS Fargate, Azure Container Apps, and Kubernetes clusters. Developed in collaboration with NVIDIA, the platform integrates NVIDIA AI Enterprise and NIM microservices to provide specialized models like NVIDIA NeMo Retriever and Llama Nemotron. This "AI factory" approach enables enterprises to scale from a few agentic workflows to thousands of critical use cases while maintaining control over their data and inference performance. The platform includes no-code development tools, real-time tracing, and enterprise connectors to bridge the gap between simple agent adoption and becoming a truly agent-native enterprise.
Deployment and Infrastructure
- On-Premises & VPC Deployment: CrewAI Factory can be deployed within a customer's own data center or a virtual private cloud (VPC).
- Cloud VPC Support: The platform supports deployment on AWS Fargate and Azure Container Apps.
- Kubernetes Orchestration: Enterprises can deploy via Helm charts on AWS EKS, Azure AKS, Google GKS, and Red Hat OpenShift.
- Serverless Scaling: A serverless, container-based architecture provides automatic scaling via resource allocation.
- High availability is maintained through persistent volumes and database replicas.
AI Factory Capabilities
- NVIDIA AI Integration: The platform integrates NVIDIA AI Enterprise microservices for specialized models and tools.
- NVIDIA NeMo Models: Customers can build and deploy agents using NVIDIA NeMo Retriever and NVIDIA Llama Nemotron.
- NVIDIA NeMo Guardrails are integrated to provide advanced reasoning and safety.
- High-Performance Inference: The integration provides secure, high-performance inference for retrieving knowledge from unstructured data.
Management and Governance
- No-Code Development: The platform includes a no-code development environment for building agents.
- Testing and Tracing: Real-time testing and tracing capabilities are provided for agentic workflows.
- Secure Authentication: Native integration with Auth0 and Microsoft Entra ID is available for enterprise-wide authentication.
- Enterprise Connectors: The platform includes enterprise-grade tools and connectors for business process automation.
Use cases
- Large enterprises deploy agents on their own VPCs to meet strict data privacy and security requirements.
- Technical teams use Kubernetes and Helm to manage and scale hundreds of agentic workflows across their own clusters.
- Developers build agents using NVIDIA NeMo Retriever and Llama Nemotron for high-accuracy knowledge retrieval from unstructured data.
Specifications
- Deployment
- Self-hosted, on-premises, or customer-managed VPC (AWS, Azure, Google Cloud)
- Platforms
- Kubernetes, Helm, Red Hat OpenShift, AWS Fargate, Azure Container Apps
- Compatibility
- NVIDIA AI Enterprise, NVIDIA NeMo, NVIDIA NIM microservices
- Integrations
- AWS RDS, Azure Database, Auth0, Microsoft Entra ID, Slack, Gmail, Salesforce
- Sign-in and administration
- Native integration with Auth0 and Microsoft Entra ID for enterprise-wide authentication
- License
- Proprietary (Enterprise license)
- Supported Cloud Providers
- AWS, Azure, Google Cloud
- Container Services
- AWS Fargate, Azure Container Apps
- Orchestration Tools
- Kubernetes, Helm, Red Hat OpenShift
- Database Compatibility
- AWS RDS, Azure Database
- Authentication Providers
- Auth0, Microsoft Entra ID
- AI Framework Integration
- NVIDIA AI Enterprise, NVIDIA NeMo, NVIDIA NIM
- Architecture Type
- Serverless, container-based
How it works
- Infrastructure Selection: Enterprises select their preferred infrastructure, such as a private data center, AWS/Azure VPC, or a Kubernetes cluster.
- Deployment and Setup: The platform is deployed using Helm charts or supported container services like AWS Fargate or Azure Container Apps.
- Configuration and Integration: Teams integrate their own cloud databases like AWS RDS or Azure Database and configure enterprise authentication via Auth0 or Microsoft Entra ID.
- Agent Orchestration: Users build, run, and monitor multi-agent workflows using the no-code development tools and NVIDIA-powered AI models.
FAQ
What is CrewAI Factory?
CrewAI Factory is a customer-managed form factor of CrewAI Enterprise that allows organizations to run the platform on their own infrastructure, such as a data center or a virtual private cloud.
Where can I deploy CrewAI Factory?
It supports AWS Fargate, Azure Container Apps, and Kubernetes clusters via Helm on AWS EKS, Azure AKS, Google GKS, and Red Hat OpenShift.
How does it work with NVIDIA?
CrewAI Factory integrates NVIDIA AI Enterprise, providing access to NVIDIA NeMo Retriever, NVIDIA Llama Nemotron, and NVIDIA NeMo Guardrails for advanced reasoning and high-performance inference.
Does it support automatic scaling?
The platform uses a serverless, container-based architecture that provides automatic scaling through resource allocation and high availability via database replicas and persistent volumes.
Added by AntSpace from public sources (crewai.com).