File: 'atlas-revenue-q4-2025-q1-2026.csv'\nRows: 18 · Columns: month, segment, mrr, customer_count, churn_rate\n\nFull data:\n2025-10, SMB, $310,000, 1,240, 4.2%\n2025-10, Mid-Market, $250,000, 187, 2.2%\n2025-10, Enterprise, $485,000, 42, 1.4%\n2025-11, SMB, $312,500, 1,250, 4.4%\n2025-11, Mid-Market, $264,500, 193, 2.1%\n2025-11, Enterprise, $518,200, 44, 1.2%\n2025-12, SMB, $311,000, 1,245, 4.7%\n2025-12, Mid-Market, $278,800, 200, 2.0%\n2025-12, Enterprise, $555,400, 46, 1.1%\n2026-01, SMB, $304,500, 1,225, 5.2%\n2026-01, Mid-Market, $293,400, 207, 1.9%\n2026-01, Enterprise, $593,800, 48, 1.0%\n2026-02, SMB, $293,200, 1,195, 6.1%\n2026-02, Mid-Market, $308,500, 213, 2.0%\n2026-02, Enterprise, $625,400, 50, 0.9%\n2026-03, SMB, $279,800, 1,160, 7.0%\n2026-03, Mid-Market, $325,200, 220, 1.8%\n2026-03, Enterprise, $665,000, 52, 0.8%\n\nAuto-detected: SMB MRR peaked in November and has declined every month since, while Mid-Market and Enterprise grew steadily. Total MRR is up across the period despite SMB decline. SMB churn rate climbed from 4.2% to 7.0%; Mid-Market and Enterprise churn stayed low and stable.

Summary  
This chapter demonstrates how to build a prompt-based agent by defining its identity, tool guidance, output format, and constraints so it reliably invokes tools and prints structured KPI briefs.  

General domain of usage  
Business analytics (weekly KPI reporting)

It's 8am Monday. You need a one-page brief on how Atlas did last week — pulled from the latest data, flagged with what changed, ready to forward to the leadership team.

You're going to build the agent that produces this brief. Your job: the persona.

## Four Moves to a Working Agent

| Move | What to Write |
|---|---|
| **Identity** | "You are a weekly KPI brief writer for a SaaS operator. You don't summarize — you flag what changed." |
| **Tool guidance** | "Always call spreadsheet_analyze before writing. Never write from memory." |
| **Output format** | "Return three blocks: Headline, 3 Movers, One Question for Leadership." |
| **Constraint** | "Never report a movement without comparing to the prior period. Never invent numbers." |

Stack the four into one prompt. That's your agent.

## A Worked Persona

><b>You are a weekly KPI brief writer for a SaaS operator. You don't summarize the data — you flag what changed and what needs attention.
>
>Always call spreadsheet_analyze before writing. Never write from memory.
>
>Return three blocks:
>- Headline — one sentence naming this week's most important shift.
>- 3 Movers — numbers that moved meaningfully, each citing the row or column they came from.
>- One Question for Leadership — a specific decision the data forces.
>
>Never report a number without comparing to the prior period. Never invent values.

Use this as a starting point, or write your own from the four parts.

>## Task
>
>Configure the agent. Then send it:
>
>><b>Run this week's KPI brief on the Atlas revenue data.
>
>If the output is mushy ("numbers look healthy"), tighten the Output Format. If it invents comparisons that aren't in the data, strengthen the Constraint. Iterate until the brief is sharp enough to forward without editing.
>
>Done when the agent produces a brief you'd actually send to a CEO.

A hands-on course for non-technical managers and operators who need to turn raw business data into decisions, not summaries. You'll learn how to frame analytical questions sharply, read spreadsheets and internal docs through AI, combine internal numbers with market context, and build your own analyst agent that produces decision-ready briefs end-to-end.

Introduction to business analytics through the lens of AI — why raw data doesn't make decisions, the three layers of analytical questions, and where an AI assistant actually fits in the workflow.



Train the framing reflex. Turn vague business questions into sharp ones, pick the metric that actually matters, and lock down time windows, segments, and comparisons before touching any data.



Hands-on labs with the three sources every operator hits: sales spreadsheets, strategy docs in Notion, and market context from the web. Single-tool labs first, then a combined investigation.



Move from running labs to building your own. Write the persona, wire up the tools, and create an agent that delivers weekly KPI briefs and pre-meeting decision packs — finishing with a capstone Quarterly Business Review.

