Discover All Cortex Agents In A Single Click: Agent Observability for Snowflake Intelligence
Table of Contents
In 2025, enterprise teams launched dozens of AI pilots. This year, they are launching thousands of agents into production…and almost no one has a way to discover and manage them.
This includes custom agents built from the ground up and agents running on managed platforms like Snowflake Intelligence. Scaling governance is not optionally–its an imperative. Without a central agent control plane, you are flying blind.
To help teams gain the visibility they need into their agentic fleet, Agent Observability is now natively integrated with Snowflake Intelligence. With a few permission grants in Snowflake and a couple of clicks in Monte Carlo’s settings, ALL of your Cortex Agents are connected — no custom instrumentation, no code changes, no SQL required.
The Monte Carlo Asset Catalogue surfaces all of your agents together, no matter where they live, allowing users to access an overview of each agent’s usage, error rate, latency and other key metrics as well as dive deep into specific traces using Trace Explorer.
Once ingested within the Monte Carlo platform, teams can monitor, triage, and resolve incidents across their entire data + AI landscape.
The gap we’re closing
Cortex Agents log detailed trace data natively to Snowflake: conversation history, tool execution, LLM planning, response generation. The data is there, but there is no object to discover. There is no registry to look up.
Without a purpose-built observability layer, accessing it means
- Knowing its there in the first place; and
- Writing SQL directly against Snowflake’s native observability tables
That meant teams had no reliable way to:
- Catch quality or performance regressions in agent outputs
- Inspect individual traces to understand agent behavior
- Monitor agents at scale with the same rigor applied to custom-built systems
For example leading data + AI team are telling us:
“We are mainly focusing on the platform and the shared capabilities…we want to have this governance and visibility on the platform level.” – AI Architect, manufacturing industry
“Teams are deploying [agents] into their services as they spawn up these workflows and it’s a bit decentralized in terms of how exactly should we be doing things?” – Software Engineer, technology industry
What’s now possible
Once connected, everything flows into the full Agent Observability experience including:
- Trace viewer: Inspect individual Cortex Agent traces and spans without writing a line of SQL. See inputs, outputs, tool calls, and the full conversation thread in a purpose-built UI.
- Conversation viewer: Review multi-turn conversations in context, including the full thread history that Cortex logs natively.
- Trace dashboards: Get aggregate visibility across all your Cortex Agent activity — volume, performance, status — in the same dashboards you use for the rest of your agent fleet.
- Agent monitors: Create any of Monte Carlo’s four agent monitor types on Cortex Agents. Catch output quality issues, flag performance regressions, and get alerted to problematic behavior.
Better Together: Snowflake Intelligence & Agent Observability
Monte Carlo already has deep integration with Snowflake across data and AI reliability, and Cortex is seeing meaningful enterprise adoption. Many of the teams already using Monte Carlo to monitor their data pipelines are the same teams now deploying Cortex Agents on top of that data.
If your team is running Cortex Agents on Snowflake Intelligence, you can connect them to Monte Carlo today. Monte Carlo is available directly on the Snowflake Marketplace and as a native app, making it easy for Snowflake customers to get started using existing Snowflake credits — no separate procurement required.
Check out the documentation for the full setup guide, or schedule a demo to learn more.
Our promise: we will show you the product.
