Claude Project: Build an S&OP Intelligence Assistant

What This Builds

A persistent Claude Project loaded with your S&OP templates, historical context, planning assumptions, and category knowledge — so every time you sit down to write your demand review, Claude already knows your process, your categories, your standard language, and what happened last month. Instead of re-explaining context every session, you start with a knowledgeable partner that remembers everything you've configured.

Prerequisites

• Claude Pro subscription ({{tool:Claude.price}}/month) — Projects require Pro
• 3–5 cycles of completed demand review content (commentary, assumptions, templates)
• 1–2 hours for initial setup

The Concept

A Claude Project is a persistent workspace where your documents and instructions are loaded permanently. Think of it as having a senior demand planning analyst who has read all your historical reviews, memorized your category structure, knows your audience's preferences, and has your templates at their fingertips. Every new conversation in the project starts from that shared understanding — no re-explaining needed.

The key capability: instead of giving Claude context every time, you do it once at setup. Then each demand review cycle, you just paste in your new numbers.

Build It Step by Step

Part 1: Gather Your Knowledge Base Documents

Collect these documents to load into the project:

1. Demand Review Template — your fill-in-the-blanks template (from Level 3 guide)
2. Category Cheat Sheet — 1-2 paragraphs per category describing the market structure, key customers, typical seasonality, and common variance drivers
3. Last 2-3 Demand Review Commentaries — so Claude learns your tone and style
4. Planning Assumptions Log — current cycle assumptions for each category (growth rates, promotional calendar, key account status)
5. Audience Guide — who attends your demand review and what each person cares about

Part 2: Create the Claude Project

1. Log in to claude.ai with your Pro account
2. Click Projects in the left sidebar
3. Click New Project
4. Name it: "S&OP Demand Review Assistant — [Company/Team]"

Part 3: Write the Project Instructions

Click Edit project instructions. This is the most important step — it defines how Claude behaves in every conversation:

You are my S&OP Demand Review writing assistant. I am a demand planner at [company type]. My planning cycle is [monthly/weekly]. I cover [X categories] across [X regions/channels].

Your role:
1. Help me write forecast variance commentaries using the template and style in the documents loaded here
2. Draft executive summaries that match the audience guide I've loaded
3. Generate category-specific commentary using the category knowledge and historical context I've provided
4. Help me structure risk/opportunity sections based on current planning context

Behavioral rules:
- Always match the tone and length conventions from my historical commentaries
- When I give you numbers, check them for internal consistency before using them (e.g., category variances should sum to total)
- For any forward-looking statement, ask me whether this is a one-time factor or structural change before committing to language
- Flag if a variance is unusually large (>15%) so I can double-check the data
- Never invent drivers or context — only use what I provide or what's in the loaded documents

When I start a new session, prompt me: "Ready to work on the [cycle] demand review. What data do you have for me today?"

Part 4: Load Your Documents

Upload each document you collected in Part 1. For text documents, paste the content directly. For your historical commentaries, paste the most recent 2–3 examples.

What you should see: Each document listed in the project's file panel. Claude now has this context for every conversation.

Part 5: Test with a Live Commentary Request

Start a new conversation in the project. When Claude says "Ready to work on the [cycle] demand review," paste your first category's data:

Category: [name]
Actual: [X]
Plan: [Y]
Variance: [%]
Top drivers:
1. [driver + reason]
2. [driver + reason]
3. [driver + reason]

What you should see: Commentary that uses YOUR language, YOUR template format, and YOUR audience's preferred style — because Claude learned it from the documents you loaded.

Part 6: Build Up Over Time

After each demand review cycle, paste the finished commentary back into the project as a new document. Over time, Claude accumulates months of your actual commentary history — making its suggestions increasingly accurate and consistent with your style.

Real Example: A Full Demand Review Session

Setup: You've loaded 3 months of past commentaries, your category cheat sheet (4 categories), your template, and your audience guide.

Session start: Claude: "Ready for the March demand review. What data do you have for me?"

You provide: Core Beverages category: 4.2M actual vs. 4.45M plan (-5.6%). Drivers: Premium line -180K (competitive entry), Value line -95K (delayed promo), Innovation +15K (new launch outperformance).

Claude produces: Commentary that matches your March template, uses language consistent with your prior months, correctly labels the competitive entry as Structural and the promo delay as Timing, and ends with the forward-look framing your VP prefers.

You provide: Next category numbers.

30 minutes later: All 4 category commentaries, executive summary, and risk/opportunity section — complete. Total time: 30 minutes vs. 3+ hours.

What to Do When It Breaks

• Commentary doesn't match your style → Add more examples of your past work to the project documents; the more samples, the better the style matching
• Claude invents context you didn't provide → Strengthen the "never invent" instruction in your project settings; add "If you need context I haven't provided, ask me rather than assuming"
• Numbers don't add up in the commentary → Add "always verify that category variances sum to total portfolio variance" to project instructions

Variations

• Simpler version: Skip the Project setup; just paste your template and last month's commentary into a regular Claude conversation each month — less elegant but still saves time
• Extended version: Add competitor activity tracking (paste monthly competitive intelligence summaries into the project) — Claude can then factor competitor context into commentary automatically

What to Do Next

• This week: Build the project; load documents; test with one category's data
• This month: Use for a full demand review; add the output back to the project history
• Advanced: Add a "category knowledge" document for each major category as your institutional knowledge grows

Advanced guide for Demand Planner professionals. These techniques use more sophisticated AI features that may require paid subscriptions.