What is AI SRE?

Observe by Snowflake AI SRE is your AI-powered site reliability engineering (SRE) observability assistant. Use AI SRE to help you investigate incidents, debug slow or error-prone services, and summarize observability insights across logs, metrics, and traces all through natural language.

AI SRE models

AI SRE uses Snowflake Cortex as the default model gateway, and also uses Cortex Embed for memories and Search for documentation. AI SRE does not call Anthropic/OpenAI directly for Cortex-routable models.

• AI SRE generates lightweight content such as titles, summaries, time-range parsing, field selection, and similar helpers using smaller models like llama3.1-70b on Cortex.
• AI SRE memories are embedded with Cortex snowflake-arctic-embed-l-v2.0 (1024-d) via inference:embed. If your local Snowflake region doesn’t support Embed REST, we send it to the closest supported region. See Model availability in the Snowflake documentation for information about Snowflake Cortex Embedding API availability.
• Documentation search uses Cortex Search when AI SRE has a need to hit Observe by Snowflake's documentation API. See Cortex search in the Snowflake documentation.

Use AI SRE in Observe by Snowflake

AI SRE is available in a variety of places in Observe by Snowflake.

You can access AI SRE by selecting AI SRE in the navigation bar, or directly from the Home page:

From the Home page, begin by selecting one of the pre-populated questions, or ask AI SRE a question directly in the text field, such as Why did latency spike last night? or Show the slowest services in my cluster. The AI SRE agent uses a range of skills and tools, also available to the MCP server and CLI, to run queries and summarize insights in the chat.

You can also use AI SRE to create and run skills and configure connectors. See Create AI SRE skills and Configure AI SRE connectors.

In addition to the AI SRE agent, many areas of the product have their own agents customized for that area:

• Trace Explorer
• APM tools (services, environments, invocation graph)
• Monitor creation (for customers with AI SRE)

Agents are able to select and use other agents as needed for whatever the user asks any place in the product, including the Home page.

For example, you can use the Monitor creation agent to help you create a Monitor from scratch by providing a description of what you want to monitor. See Threshold Monitor example: Check for CrashLoopBackOff for an example of how to start building a Monitor using AI SRE.

Or, during the Monitor creation workflow, where you can see the Monitor-specific context already provided:

Legal notices

The data classification of inputs and outputs are defined as follows: