---
name: al-695-answer-analytics-questions-with-a-governed-multi-agent-as
description: "phData's Nov 2026 case study describes a two-person pod that put a governed multi-agent analytics assistant into production in 10 weeks for a $9B+ enterprise software company (~21,000 employees, ~150 certified data products). A Claude Opus orchestrator splits each natural-language question across specialized sub-agents on Strands and Amazon Bedrock AgentCore (Sonnet for heavy reasoning, Haiku for faster analysis). Answers come only from governed Snowflake semantic views and Atlan metadata via MCP—not free-form SQL on raw tables—with Okta user-scoped tokens and read-only access. A Golden Query Bank of 100+ verified queries gates releases (~93% governed-metric grounding vs a 90% threshold). Result: analysis turnaround fell from about 1 hour to about 15 minutes (75%). Use when the user wants help with governed multi-agent analytics on semantic metrics (research) for a small business."
---

# Answer analytics questions with a governed multi-agent assistant: orchestrate retrieval, analysis and reporting on approved metrics only, under each user's own permissions

phData's Nov 2026 case study describes a two-person pod that put a governed multi-agent analytics assistant into production in 10 weeks for a $9B+ enterprise software company (~21,000 employees, ~150 certified data products). A Claude Opus orchestrator splits each natural-language question across specialized sub-agents on Strands and Amazon Bedrock AgentCore (Sonnet for heavy reasoning, Haiku for faster analysis). Answers come only from governed Snowflake semantic views and Atlan metadata via MCP—not free-form SQL on raw tables—with Okta user-scoped tokens and read-only access. A Golden Query Bank of 100+ verified queries gates releases (~93% governed-metric grounding vs a 90% threshold). Result: analysis turnaround fell from about 1 hour to about 15 minutes (75%).

## Inputs to collect first
Ask the user for anything below that they haven't already given you. Don't guess missing facts.

- Team
- E.g. CX / sales funnel / product usage
- Snowflake semantic views, Atlan, Okta, warehouse
- Claude via Bedrock or similar

## Instructions

Do the task below with the user's inputs in place of each [bracket].

Design a governed multi-agent analytics assistant for [team]. Question types to start: [e.g. CX / sales funnel / product usage]. Data stack: [Snowflake semantic views, Atlan, Okta, warehouse]. Models: [Claude via Bedrock or similar]. Hard rules: no free-form SQL on raw tables; every metric must resolve to a governed definition; read-only; queries run as the requesting user; ship only if eval accuracy ≥ [90%]. 1) Map orchestrator + sub-agents (retrieve, analyze, chart/report, insight). 2) List MCP tools and permission boundaries. 3) Sketch a 20-query Golden Query Bank and pass/fail rubric. 4) Draft a 10-week build plan with the go-live accuracy gate. Do not run live warehouse queries here.

## Safety checks (share these with the user)

- Check every figure, price and total against your own records before anything is sent, posted or booked. A person approves refunds, discounts, payments and price changes.
- Remove names, contact details, account numbers and other personal data the task doesn't need before you paste, and check your AI tool's data and training settings.
- Check names, dates, numbers and any claim against your own source material before you use the answer. AI can misread or invent details.
- Say clearly which parts need a person to review or approve before anything is sent, paid, posted or filed.

## About this skill

- Works with: Claude, Amazon Bedrock AgentCore, Strands Agents, Snowflake, Atlan, Okta
- Based on this real-world example: https://www.phdata.io/case-studies/governed-multi-agent-analytics-assistant/
- Free prompt from AgentPlaybook Real Estate: https://agentic-library.netlify.app/prompt/al-695-answer-analytics-questions-with-a-governed-multi-agent-assis/
- To install in an agent that supports Agent Skills, save this file as SKILL.md in a folder with the same name as `name` above (for example in your skills directory). Install steps for Claude.ai, Claude Code, Cursor and Codex: https://agentic-library.netlify.app/skills/