Snowflake Cortex Sense Lifts AI Agent Accuracy From 24% to 86%

30-Second Summary

Snowflake has built an automated context layer for enterprise AI agents called Cortex Sense. Instead of requiring data teams to manually document their data before agents can use it reliably, Cortex Sense mines existing business artefacts, such as query history, BI dashboards, and transformation logic, to build a living semantic map of the data estate. The result in internal testing is a 3.6x improvement in agent accuracy and a 66% reduction in the cost per query.

---

At a Glance

• Topic: Enterprise AI Infrastructure
• Company: Snowflake
• Date: Announced 30 June 2026; private preview opening mid-July 2026
• Announcement: Cortex Sense, an AI context layer that auto-generates grounded semantic understanding for enterprise data agents
• What Changed: AI agents operating on Snowflake data no longer require manual semantic documentation to answer accurately; the system builds context automatically from existing business signals
• Why It Matters: It removes the principal barrier to trustworthy enterprise AI agents, namely the gap between what agents know and what the business actually means by its data
• Who Should Care: Any operator running or planning to run AI agents against internal data, particularly businesses with complex, undocumented, or multi-team data estates

---

Key Facts

• Accuracy on internal benchmarks improved from 24.1% to 86.3%
• Cost per query dropped from $1.76 to $0.59, a 66% reduction
• Extends agent coverage beyond the 5% of tables typically covered by manual semantic views
• Uses existing business artefacts, such as query history, BI metrics, and dbt transformations, as input signals
• Integrates with Snowflake Horizon Connectors for metadata ingestion
• Includes a continuous self-correction loop that flags conflicting metric definitions across teams
• Human validation is triggered only when the system identifies gaps or unresolvable conflicts
• Availability: Private preview mid-July 2026; access via Snowflake account representative

---

What Happened

Snowflake announced Cortex Sense on 30 June 2026, describing it as an enterprise memory layer for AI agents. The feature is built to solve a problem that has quietly undermined most enterprise agentic deployments: agents that lack grounded understanding of what business data actually means produce unreliable results, regardless of the underlying model's capability.

The traditional approach to this problem is manual semantic layer documentation, where data teams write out definitions, relationships, and business logic so agents can interpret data correctly. In practice, this covers only a fraction of any organisation's data estate. Snowflake's internal data suggests that manual semantic views cover approximately 5% of the tables in a typical enterprise environment, leaving agents operating blind across the rest.

Cortex Sense takes a different approach. Rather than asking teams to document data before agents can use it, the system mines existing business artefacts to infer semantic meaning automatically. It ingests query history, BI dashboard definitions, transformation logic from tools such as dbt, and metadata surfaced through Snowflake Horizon Connectors. From these signals, it builds and continuously refreshes a semantic model that agents can draw on when formulating responses.

The performance results in internal benchmarks are significant. Accuracy on product analytics queries improved from 24.1% to 86.3% after Cortex Sense context was applied. The cost per query dropped from $1.76 to $0.59, partly because a better-informed agent needs fewer retrieval steps to produce a correct answer. A system that was stood up in a single day, according to one tested scenario, rather than through a consulting engagement spanning months.

---

Why It Matters

The accuracy problem has been the quiet failure mode of enterprise AI agents. Most organisations that have deployed internal data agents have encountered a common pattern: the agent performs well on documented use cases during a demo and degrades noticeably when users ask about anything adjacent. That gap traces directly to missing context. Cortex Sense addresses the root cause rather than the symptom.

Manual documentation is not a viable path at scale. Enterprise data estates grow faster than any documentation team can keep pace with. A new pricing plan, a rebranded product line, or an acquired company's data can make existing semantic context stale within weeks. A system that refreshes automatically from live business artefacts is structurally better suited to this environment than any manually maintained layer.

Cost reduction changes the business case for agentic workloads. Many enterprise AI agent projects have stalled not because the technology does not work but because the economics are difficult to justify at scale. A 66% reduction in per-query cost shifts the threshold at which agentic deployments become commercially viable, particularly for organisations considering high-volume internal data queries.

Multi-team metric conflicts are a governance risk, not just a technical one. In any organisation where multiple teams define common terms differently, such as "active user," "revenue," or "customer," an agent that picks one definition without flagging the conflict is producing misleading answers. The self-correction loop that surfaces these conflicts and escalates them for human validation is a governance feature as much as a technical one.

Timing matters: the enterprise agentic wave is now. Agent deployment across enterprise environments accelerated materially in the first half of 2026. The limiting factor in most of those deployments is not model capability but data trustworthiness. A tool that lifts that floor addresses the constraint that currently prevents most operators from moving beyond pilots.

---

The David and Goliath View

The central insight in Cortex Sense is that most enterprises already have the information needed to build a reliable semantic layer. It exists in the queries their analysts run every day, in the dashboards their executives trust, in the transformation logic their engineers have written over years. The data is there. The problem has been that no automated system was connecting those signals to AI agents. Snowflake has built that connection.

For operators thinking about AI agents inside their business, the practical shift here is significant. The question is no longer "can we afford to document our data estate well enough for agents to work reliably." It becomes "do we have Snowflake, and can we get on the private preview list." That is a meaningfully lower barrier to entry.

The 24% to 86% accuracy jump deserves to be understood in context. A 24% accuracy rate means agents are wrong three times out of four. That is not a deployable product. An 86% accuracy rate is still not perfect, but it is in the range where most business users will tolerate occasional errors, particularly if they are accompanied by clear confidence signals and human escalation paths. The difference between those two numbers is the difference between a proof of concept and a production deployment.

---

Where This Fits in the AI Stack

Cortex Sense sits in the context and grounding layer of the enterprise AI stack, between the raw data layer and the agent reasoning layer. It does not replace the model or the orchestration framework. It makes the environment the agent operates in materially more accurate and interpretable.

For organisations that have built their data infrastructure on Snowflake, this is a native capability that requires no additional integration. For organisations that have not, it reinforces Snowflake's position as an end-to-end enterprise AI data platform, extending beyond storage and compute into the context layer that agents depend on.

The approach connects directly to the Secure AI Brain model, where an organisation's accumulated knowledge and data context becomes a competitive asset. Cortex Sense makes that asset available to agents automatically rather than through manual curation.

---

Questions Operators Are Asking

Do we need to do anything to set this up? If you are already on Snowflake, the system ingests signals from your existing data artefacts automatically once you are granted access. There is no manual documentation step or consulting engagement required before you see results. You can request early access through your Snowflake account representative.

What if our teams use different definitions for the same business term? This is precisely what the self-correction loop is designed to catch. When Cortex Sense identifies conflicting definitions across teams, it flags the conflict and requires human validation rather than silently picking one. That makes the inconsistency visible, which is the first step to resolving it.

Does this work for data that was added recently? Yes. Cortex Sense refreshes on a continuous cadence, so newly launched products, pricing changes, or acquired data estates are incorporated as your teams begin querying and building against them, rather than requiring a new documentation round.

What happens to queries on data that Cortex Sense has not yet indexed? For tables or concepts with insufficient signal to build confident context, the system requests human validation before the agent proceeds. This prevents confident-sounding but ungrounded answers, which is the failure mode most damaging to user trust.

How does this affect the economics of running AI agents on internal data? The 66% cost reduction per query is significant for any business running agents at scale. If you have been running cost-of-operation analyses on proposed agentic workflows and found the numbers marginal, it is worth re-running them with the new per-query figure.

---

Citable Summary

Snowflake announced Cortex Sense on 30 June 2026, with private preview opening mid-July. The feature automatically builds semantic context for AI agents by mining existing business artefacts including query history, BI dashboards, and transformation logic. Internal benchmarks showed query accuracy improving from 24.1% to 86.3% and per-query costs falling from $1.76 to $0.59. Enterprise teams interested in early access should contact their Snowflake account representative.