Scalr AI FAQ

General

Q: What AI product is being used?
A: Scalr uses Claude Sonnet, but other models will be added in the future. Administrators can choose which one to use.

Q: Is it a shared or dedicated model?
A: Shared, but with 0 days retention policy, meaning the data is only used to generate the answer and not stored or used for anything else.

Q: Is the data used to improve the model?
A: No, data is used only to generate the answer and not stored or used for anything else.

Q: Is Scalr AI enabled by default?
A: No, the AI is disabled by default and can be enabled in the Scalr integrations page.

Q: Can I use my own (corporate) account for Anthropic, ChatGPT, etc?
A: Right now, no, but BYO credentials for the AI account are planned in the future.

Data & Privacy

Q: What information is being shared with AI?
A: As of today only plan output is used to generate help with errors or summaries for approvals.

Q: Is my infrastructure configuration data being stored or used to train the AI?
A: No, infrastructure configuration is neither stored nor used to train the AI.

Q: How is my sensitive data protected when using the AI troubleshooting feature?
A: Scalr only sends sanitized data for analysis.

Q: Does the AI feature comply with GDPR, CCPA, and other privacy regulations?
A: Yes, see how Claude handles data here.

Functionality

Q: What types of Terraform errors can the AI help troubleshoot?
A: It will handle any type of Terraform, Tofu, or custom hook error.

Q: Can the AI suggest fixes or just identify problems?
A: Yes, the goal is that the AI will help end users fix the problem themselves without having to contact the platform team.

Q: Will the AI automatically implement fixes or only provide recommendations?
A: Scalr AI only provides recommendations; it does NOT implement the fix.

Implementation

Q: How do I enable or disable the AI troubleshooting feature?
A: Scalr AI can be enabled in the integration page in Scalr adminstration.

Q: Is there an additional cost for using the AI troubleshooting feature?
A: No, this is included in the cost of your subscription.

Q: Does using the AI feature impact performance or cause delays in my workflow?
A: No, AI can only be used to troubleshoot errored runs. It does not impact in-progress runs.

Q: Will the AI feature work with custom Terraform or OpenTofu providers or modules?
A: Yes, the AI will help troubleshoot any error in a failed run.