Import and connect Jev applications
Jev is TypeSafe's model service. Import the application into its declared host, then configure its model connection separately.
Choose the import for your product
- Source or applicationDownload the purchased repository/archive from your Library, install its dependencies and configure your backend.
- Workflow or agentImport the purchased workflow into its declared host. Use a server-side HTTP step or a custom node when the host has no native TypeSafe connector.
- Browser extensionInstall the declared build and connect its backend. Keep the TypeSafe key in the backend, outside the extension bundle.
- Evaluation kitImport its permitted test data, questions and rubric; obtain TypeSafe access separately.
- Made using JevA creator's authoring declaration alone requires no runtime connection.
Set up your TypeSafe connection
- Open setup
Open the purchased installation's native preflight. The Jev setup panel checks the current version and declaration.
- Download the setup files
Download all five setup files to one private directory: the TypeSafe client, connection check, model configuration, package file and instructions.
- Add your key
Install the package dependencies. Set the configured API key variable in your host's secret store; never enter the key in Sorla or exported product files.
- Run the connection check
Run the supplied connection check. It queries the models available to your account, without running inference or sending business data.
- Use it with care
Use named questions and typed answers in your application. Send uncertain decisions for human review and keep actions disabled until separately tested and authorized.
Provider access and charges are separate from the product purchase. A passing model-access check does not verify application accuracy or enable provider actions.