On-premise deployment
Reference architecture for CODEXIS AI Agent in customer infrastructure, data storage and connections to external services.
This page is for IT teams preparing to deploy CODEXIS AI Agent in their own data centre or cloud. It describes the components and data flows you can use to plan infrastructure and network rules.
Reference architecture
CODEXIS AI Agent, user virtual computers and persistent storage run in the customer's environment. Each user has a separate workspace connected to their own data. External services are reached through controlled network connections.
The diagram shows the main layers and their connections. Service placement, authentication integration and connected services are specified during deployment design.
User environments
Users access CODEXIS AI Agent over a secure connection. Their tasks run in a dedicated virtual computer where the agent works with files and runs tools. User environments are isolated from one another.
Data storage and backups
Working files and agent context are stored on the user virtual computer's persistent home disk. The disk remains on its storage node when the virtual computer stops. A backup of the home disk and the information needed to restore it are stored in a separate backup repository.
For an on-premise deployment, both primary storage and backups reside in your environment. Backups cover the data and configuration needed to restore the environment. Define recovery time, acceptable data loss and backup retention when planning operations, and verify them with a restore test.
External services and network traffic
AI requests are processed through the OpenAI API. CODEXIS AI Agent sends the request and the context needed to answer it, which may include the contents of documents being processed. Assess this data flow alongside the location of stored customer data.
The CODEXIS API supplies legal data to the features that use it. Other connections depend on enabled integrations, such as Microsoft 365 or email. Include their destinations and transferred data in the network design. Configure access to user environments and outbound connections according to your organisation's security rules.
Preparing a deployment
We size compute nodes and storage according to concurrent users, document sizes and the tools in use. Also prepare your requirements for authentication, availability, monitoring, updates and data recovery. We agree the configuration and division of operational responsibility with your IT team. Detailed architecture and deployment documentation are provided under a non-disclosure agreement (NDA).
Read Security and data protection for data processing terms. Contact us to design a deployment for your environment.