Bottom line: Anthropic uses Claude Tag to make the data analytics foundation from Claude Code described in the previous blog post (semantic layer, skill files, evaluation suite) available for self-service queries in Slack.
In a blog post, Anthropic describes how the internal data analytics foundation from Claude Code was transferred to Slack. With Claude Tag (Public Beta), employees outside the data team can ask ad-hoc questions about company data directly in Slack and receive answers based on the same governance-compliant definitions used by analysts.
In a previous post, Anthropic described how Claude can answer data analytics questions with roughly 95 percent accuracy through three core building blocks: a governance-compliant semantic layer, a collection of skill files that encode analytical conventions, and an evaluation suite for measuring performance. That earlier post focused on Claude Code as the primary development environment for data scientists and data engineers at Anthropic, as well as on best practices for improving agentic accuracy.
The new post describes how Anthropic’s data team is transferring this foundation to the everyday work of the rest of the company. To do so, Anthropic relies on Claude Tag, currently in a public beta version, as the basis for a data analytics agent in Slack. Anyone in the company can ask this agent data-related questions and receives answers based on the same governance-compliant definitions that analysts also use internally.
For practitioners, the relevance lies less in pure model accuracy than in the deployment perspective: Anthropic emphasizes that an agent’s accuracy and its successful deployment for non-analysts represent two distinct challenges. The accuracy recommendations from the previous post therefore remain valid, but are not repeated in the current post. Instead, Anthropic focuses on five key lessons learned over the past year regarding the deployment of a data analytics agent in Slack, centered on distribution, permissions, data freshness, and observability.
For teams planning their own internal analytics agents, the post thus provides a practical starting point: the technical foundation of semantic layer, skills, and evaluation appears to be reusable, while organizational embedding into a collaboration tool like Slack brings its own requirements for access control and monitoring.
Source: claude.com · Published August 12, 2026
Lumi AI News — AI-assisted curation in accordance with Art. 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.8.3.