FAQ

​

What is AI SRE?

AI SRE is an AI-powered incident response assistant that helps support engineers, production engineers, on-call engineers, and SREs investigate and resolve production incidents faster.

​

How does it work?

Ask questions about incidents in natural language. AI SRE gathers evidence from your tools, correlates data, and provides evidence-based insights with confidence levels.

​

What do I need?

• GitHub integration (required)
• Access to the AI SRE web application
• (Optional) Observability tools such as Datadog

​

How do I enable gated connectors?

Contact support with your requirements. Support will enable the connector for your account and guide you through setup.

​

How accurate is it?

AI SRE provides confidence levels:

• High — Strong evidence, likely accurate
• Medium — Good evidence, verify
• Low — Hypothesis, needs more data

Always verify critical findings.

​

Can it make changes to my systems?

No. AI SRE is investigative only. It provides analysis and recommendations, but you make decisions and implement fixes.

​

Is my data secure?

Yes. AI SRE uses enterprise-grade security with tenant isolation, encryption, and read-only integrations. See the Security & Privacy section below for details.

​

Security & Privacy

​

Tenant isolation

• Logical controls — Tenant identifiers and API gateway-level request filtering
• Storage isolation — Dedicated storage volumes and tenant-specific encryption keys
• AI agent isolation — AI agent sandboxes are fully isolated per tenant

​

Integrations

• Read-only access — Integrations configured without write permissions

​

Data and privacy

• No write access — Customer data remains unchanged
• Least-privilege access — Permissions scoped to minimum required
• Data retention — 14 days by default (configurable)

​

Source code protection

• Read-only analysis — Real-time analysis with no code storage
• Code snippets — Only small snippets from investigations are saved (access controlled via customer-managed OAuth tokens)
• No code execution — Code execution not permitted in AI agent environment

​

AI providers and data handling

• Providers — OpenAI, Google Cloud Gemini, AWS Bedrock
• Data protections — Zero-retention contractual terms, Data Processing Agreements (DPAs), regional endpoints, prompt sanitization
• No training — Opt-out from training; LLM providers do not train on customer inputs beyond request processing time

​

How do I get help?

Contact your organization’s AI SRE administrator or support team.