# What Okta learned governing context for AI agents at scale

DevFeed: [What Okta learned governing context for AI agents at scale](<https://devfeed.tech/articles/what-okta-learned-governing-context-for-ai-agents-at-scale-101841.md>)

Original publisher: [Read original article](<https://www.getdbt.com/blog/okta-ai-agent-context-governance>)

Author: Daniel Poppy

Published: 2026-10-26T22:14:00Z

Content type: news

Language: en

Sources: [dbt Labs](<https://devfeed.tech/sources/dbt-labs.md>)

Topics: [AI-Ready Data](<https://devfeed.tech/topics/ai-ready-data.md>), [context](<https://devfeed.tech/topics/context.md>), [Data Infrastructure](<https://devfeed.tech/topics/data-infrastructure.md>), [data-architecture](<https://devfeed.tech/topics/data-architecture.md>), [business logic](<https://devfeed.tech/topics/business-logic.md>), [okta](<https://devfeed.tech/topics/okta.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [context](<https://devfeed.tech/tags/context.md>), [context-for-ai-agents](<https://devfeed.tech/tags/context-for-ai-agents.md>), [dbt](<https://devfeed.tech/tags/dbt.md>), [learn](<https://devfeed.tech/tags/learn.md>), [okta](<https://devfeed.tech/tags/okta.md>), [scale](<https://devfeed.tech/tags/scale.md>), [source](<https://devfeed.tech/tags/source.md>), [source-of-truth](<https://devfeed.tech/tags/source-of-truth.md>)

## AI overview

Okta found that governing what AI agents can see and trust was harder than building the agents. Its early Snowflake semantic views were disconnected from dbt's business logic, lacked an external catalog and auditability, and bundled instructions without distinguishing stable knowledge from information to look up. The team moved governance into a three-layer dbt framework: model and column metadata for safe uses, exclusions and automatic filters; Markdown docs for company facts and domain routing; and PR-reviewed semantic views for facts, dimensions and metrics. A custom skill converts existing Snowflake semantic-view DDL into dbt specs. The layers sync to Agents Schema: Claude reads metadata over MCP and docs as Markdown, while Cortex uses a search service and its own skill configuration. Both therefore use the same context through different interfaces, and a merged dbt change reaches consuming agents without separate config edits. Planned work includes regression tests, broader domain coverage and feedback from agent answers into context docs; these are future plans, not reported capabilities.

## Source excerpt

Okta rebuilt how it governs context for AI agents, using dbt as the single source of truth.