Learn AI-PHISR

A hands-on course to build your own PM operating system. No programming required, but you will copy and paste a few terminal commands. You'll type, see results, and learn by doing.

8 Lessons ~90 minutes Copy/Paste Friendly Your Own Product by End

What You'll Build

By the end of this course, you'll have a fully working AI-powered PM system for your actual product. The AI will know your strategy, your team, your initiatives, and your decisions — and it will remember all of it across every conversation.

Who is this for? Product managers who want an AI assistant that actually understands their product context. You do not need to know how to program. If you can fill out a worksheet, open Obsidian, and paste a few terminal commands, you have all the skills you need.
Important The terminal screenshots in this course are examples, not exact guarantees. Successful behavior matters more than matching every line word-for-word.

Lesson Map

  1. What is AI-PHISR? — Understand the concept (10 min)
  2. Install the Tools — Get everything running (15 min)
  3. Your First Conversation — Talk to Claude Code (10 min)
  4. Fill Your Setup Worksheet — Tell the system about your product (15 min)
  5. Bootstrap Your Vault — One command to populate everything (5 min)
  6. Run Your First Skill — Build a hypothesis from a real idea (10 min)
  7. Run Your First Playbook — Full multi-step workflow (10 min)
  8. Your Daily Workflow — How to use this every day (15 min)
1

What is AI-PHISR?

Before we install anything, let's understand what we're building and why it matters.

The Problem

Every time you use ChatGPT, Claude, or any AI tool, you start from zero context. You have to re-explain your product, your strategy, your team, and your current situation. Every. Single. Time.

Imagine if you had a PM assistant who already knew everything about your product — your strategic bets, your active initiatives, your open decisions, your team dynamics. And it remembered what you discussed yesterday.

The Solution

AI-PHISR is that assistant. It works by combining two things:

📚
Obsidian

A free note-taking app. Your product knowledge lives here as simple text files.

⌨
Claude Code

An AI tool you type to in a terminal. It reads your notes and gives smart, contextual answers.

The magic: Claude Code reads your Obsidian notes before answering. So when you ask "what should I focus on today?", it already knows your sprint goals, your blocked items, and your open decisions.

What's a Terminal?

A terminal (also called "command line" or "command prompt") is a text-based way to interact with your computer. Instead of clicking icons, you type commands. It looks like this:

Terminal
$ claude Welcome to Claude Code! Type your message to get started. You: What should I focus on today?

That's it. You type, Claude responds. No code to write, no programming language to learn. Just plain English (or any language you prefer).

Key Takeaway AI-PHISR = your product knowledge (in Obsidian) + an AI that reads it (Claude Code). The AI gets smarter as you add more knowledge. That's the whole idea.
2

Install the Tools

Let's get everything installed. This takes about 15 minutes. Follow each step carefully.

Step 1: Get a Claude Subscription

You need a Claude Pro ($20/month) or Max ($100/month) subscription. This gives you access to Claude Code.

  1. Go to claude.ai in your web browser
  2. Create an account (or sign in if you have one)
  3. Click Upgrade and choose Pro or Max

Step 2: Install Node.js

Node.js is a small program that lets Claude Code run on your computer. You only need to install it once.

  1. Go to nodejs.org in your browser
  2. Click the big green button that says "LTS" (Long Term Support)
  3. Run the downloaded file and click Next through all the steps
  4. When it's done, open a terminal to verify it worked

How to open a terminal:

In the terminal, type this and press Enter:

Terminal
PS> node --version v20.11.0 PS> npm --version 10.8.2
What you should see A version number like v20.11.0 for Node and another version number for npm. Your numbers may be different. If you see an error like "not recognized" or "command not found", close the terminal, reopen it, and try again. If it still doesn't work, reinstall Node.js.

Step 3: Install Claude Code

Type this in your terminal npm install -g @anthropic-ai/claude-code
Terminal
PS> npm install -g @anthropic-ai/claude-code added 1 package in 15s PS> claude --version claude-code v1.x.x
What you should see A message saying packages were installed, and then a version number when you run claude --version. This means Claude Code is ready.
Troubleshooting
  • "EACCES permission denied" (Mac/Linux) — Run sudo npm install -g @anthropic-ai/claude-code instead. Enter your computer password when prompted.
  • "npm is not recognized" — Close the terminal, reopen it, and try again. If it persists, reinstall Node.js and make sure to check "Add to PATH" during install.
  • "claude is not recognized" (Windows) — Close and reopen PowerShell after the install. If it still fails, run the install again and confirm npm finished successfully.
  • Slow install? — It can take 1-2 minutes on slower connections. Wait for it to finish.

Step 4: Install Obsidian

  1. Go to obsidian.md in your browser
  2. Click Download (it's free)
  3. Install and open Obsidian
  4. When asked, click "Open folder as vault"
  5. Navigate to the ai-phisr folder, then select the vault folder inside it
What you should see in Obsidian A left sidebar with folders: Strategy, Initiatives, Hypotheses, Decisions, Research, Team, Sprints, System. If you see these, Obsidian is set up correctly.
Obsidian — ai-phisr vault
▼ Strategy
vision.md
strategic-bets.md
okrs.md
▶ Initiatives
▶ Hypotheses
▶ Decisions
▶ Research
▶ Team
▶ Sprints
▶ System

# Product Vision

---
title: (your product name here)
---

Welcome to your AI-PHISR vault.
Start by filling out the Setup Worksheet.
Checkpoint You now have: Claude subscription + Node.js + Claude Code + Obsidian with the vault open. You're ready for the fun part.
3

Your First Conversation

Let's start Claude Code and have your first conversation. This is where it starts feeling magical.

Navigate to the AI-PHISR Folder

In your terminal, you need to go to the ai-phisr folder. Type this and adjust the path to where you put the folder:

Terminal
# On Windows PowerShell: PS> cd "C:\DEV\local\ai-phisr" PS> Get-Location Path ---- C:\DEV\local\ai-phisr PS> dir CLAUDE.md docs vault Setup_Worksheet.md # On Mac/Linux: $ cd ~/Documents/ai-phisr
Important The cd command means "change directory" — it tells the terminal to go to a specific folder. You must be inside the ai-phisr folder for everything to work, because that's where the CLAUDE.md file lives that tells Claude how to behave. On Windows, keep the quotes if your path has spaces.

Start Claude Code

Type this in your terminal claude
Terminal — Claude Code
PS> claude ━━━ Claude Code ━━━ Startup text may vary by Claude Code version Type your message, or use / for slash commands You: █
What you should see Claude Code starts and shows that it is ready for input. The exact startup text may differ, but successful behavior means: Claude opens in the ai-phisr folder, accepts input, and lets you use slash commands.
Troubleshooting
  • "claude: command not found" or "claude is not recognized" — Close and reopen your terminal, then try again. If it still fails, reinstall with npm install -g @anthropic-ai/claude-code.
  • Claude asks you to sign in — Follow the on-screen instructions. It will open a browser window for authentication. This only happens once.
  • Startup text looks different from this page? — That's normal. Make sure you ran cd into the ai-phisr folder first and that Claude accepts input.
  • Not sure you are in the right folder? — Run Get-Location and dir in PowerShell. You should see CLAUDE.md and vault/.

Say Hello

Let's have a simple conversation to see how it works. Type this exactly:

Type this at the prompt Hi! What is AI-PHISR and what can you help me with?
Terminal — Claude Code
You: Hi! What is AI-PHISR and what can you help me with? I'm AI-PHISR, your PM operating system. I maintain persistent context about your product across all sessions. Here's what I can help with: Skills (type / to see the full list): /hypothesis-builder - Frame testable hypotheses /1on1-prep - Prep for manager 1:1 /prd-builder - Generate a PRD /priority-stack - Prioritize your work ... and 17 more Playbooks (multi-step workflows): /playbook-new-idea - Process new requests /playbook-sprint-kickoff - Start a sprint /playbook-manager-1on1 - Prep for 1:1 ... and 5 more To get started, fill out the Setup_Worksheet.md file, then run /populate-strategy to bootstrap your vault.

Claude already knows what it is and what skills are available because it read the CLAUDE.md file in this repo. Now let's try the slash command menu:

Type this / (just the forward slash, then look at what appears)
What you should see A list of all available slash commands (skills and playbooks) will appear. You can scroll through them or start typing to filter. Press Escape to close the menu without running anything.
Where slash commands come from Slash commands are defined in .claude/skills/. The command name comes from each file's command: field, so the filename and slash command do not always match exactly. For example, one-on-one-prep.md defines /1on1-prep.
Checkpoint You just had your first conversation with AI-PHISR! You can see it knows about skills, playbooks, and the setup process. Now let's give it some real product context.
4

Fill Your Setup Worksheet

This is the most important step. You're going to tell AI-PHISR about your actual product. The better you fill this out, the smarter the AI will be.

Open the Worksheet

Find the file called Setup_Worksheet.md in the ai-phisr folder. You can open it with:

Fill Out Each Section

The worksheet has clear prompts. Here's an example of how to fill it out. Let's say your product is a food delivery app called "QuickBite":

Setup_Worksheet.md
## Product Overview

Product Name: QuickBite

One-line Description:
A food delivery app that guarantees 15-minute delivery for lunch orders in urban areas.

Target Users:
Office workers in downtown areas who want fast, affordable lunch without leaving their desk.

Key Problems Solved:
1. Lunch delivery takes 30-45 min, too long for a work break
2. Delivery fees make ordering lunch daily too expensive
3. Menu options are overwhelming — too many choices, not enough curation

## Strategy

Product Vision:
Make ordering lunch as fast and easy as grabbing a coffee.

Bet 1:
- Name: Micro-kitchen network
- Thesis: Partnering with small kitchens within 1km of office
clusters will enable 15-min delivery
- Success looks like: 90% of orders delivered in under 15 min
- Time horizon: 6 months
Don't overthink it You don't need to fill every single field perfectly. The system can work with incomplete information — it will flag what's missing and suggest what to add later. But the more you fill in now, the better the AI will understand your product from day one.

Sections to Prioritize

If you're short on time, focus on these four sections (they have the biggest impact):

  1. Product Overview — name, description, target users, problems
  2. Strategic Bets — your 1-3 biggest bets right now
  3. Active Initiatives — what you're building right now
  4. Your Manager — name, role, and what they care about
Exercise

Fill Out Your Worksheet Now

Open Setup_Worksheet.md and fill out at least the four priority sections above using your real product information. Take 10-15 minutes.

  1. Open the file in any text editor
  2. Replace the placeholder text with your product details
  3. Save the file when done
  4. Come back here for the next step
5

Bootstrap Your Vault

Now comes the magic moment. One command will read your worksheet and populate the core strategy, initiative, team, and memory files in your vault with structured product knowledge.

Run the Bootstrap Command

Make sure Claude Code is running (if you closed it, type claude in your terminal from the ai-phisr folder). Then type:

Type this at the Claude Code prompt /populate-strategy
Terminal — Claude Code
You: /populate-strategy Reading Setup_Worksheet.md... ✔ Created vault/Strategy/vision.md ✔ Created vault/Strategy/strategic-bets.md ✔ Created vault/Strategy/okrs.md ✔ Created vault/Initiatives/_active.md ✔ Created vault/Initiatives/micro-kitchen-network.md ✔ Created vault/Initiatives/curated-lunch-menu.md ✔ Updated vault/Team/stakeholders.md ✔ Updated vault/System/memory/context-index.md Vault populated! Here's a summary: - Vision: "Make ordering lunch as fast as grabbing a coffee" - Strategic bets: 2 (Micro-kitchen network, Curated menus) - Initiatives: 2 active - Stakeholders: 3 mapped ⚠ Missing sections in your worksheet: - OKRs (no key results defined) - Cross-team dependencies - Current sprint info You can add these later. The system works without them.
What you should see A list of files created and updated, with green checkmarks or a similar summary. Claude may also tell you which worksheet sections were left empty — that's fine, you can fill them in later.
Your output will look different The example above uses the "QuickBite" demo product. Your output will show your own product name, bets, and initiatives. The file names and structure will match, but the content will reflect what you wrote in your worksheet. Claude may also ask follow-up questions or suggest improvements — that's normal and helpful.

Check Obsidian

Switch to Obsidian. You should see your vault folders now have content!

End-to-end example The flow here is: fill Setup_Worksheet.md → run /populate-strategy → Claude creates or updates files in vault/ → you open those files in Obsidian and review them. This review step matters because the vault becomes your source of truth.
Checkpoint Your vault is now populated with your product's strategic context. Every future interaction with Claude will be grounded in this data. Let's test it.
6

Run Your First Skill

Skills are single-step commands that produce one specific output. Let's build a hypothesis from a real product idea.

Scenario

Imagine someone on your team (or a stakeholder) comes to you with an idea. Maybe it's:

Think of a real idea someone has suggested for your product. We'll use it now.

Run the Hypothesis Builder

Type this at the Claude Code prompt /hypothesis-builder

Claude will ask you about the idea. Here's an illustrative conversation showing the kind of output to expect:

Terminal — Claude Code
You: /hypothesis-builder Loading context: strategic bets, active initiatives, sprint state... What's the idea, observation, or problem you want to turn into a hypothesis? You: Our sales team says enterprise customers want a subscription plan instead of paying per order. They think it would increase retention in the enterprise segment. This connects to your "Micro-kitchen network" bet — enterprise offices are a key segment. ## Hypothesis: Enterprise Subscription Plan We believe that offering a monthly subscription plan (unlimited orders, flat fee) for enterprise office teams of 20+ employees will result in 40% higher 90-day retention vs pay-per-order because subscription commitment reduces the daily decision friction and makes QuickBite a default lunch habit. ## Assumptions (ranked by risk) | Assumption | Uncertainty | Impact | Risk | |-----------------------------------|------------ |--------|------| | Enterprises will pay flat monthly | 4 | 5 | 20 | | 20+ employees actually use it | 3 | 4 | 12 | | Unit economics work at flat rate | 4 | 3 | 12 | | Subscription reduces churn | 2 | 4 | 8 | ## Success Criteria Metric: 90-day retention rate for subscribed teams Target: 40% higher than pay-per-order baseline Timeline: 8-week pilot with 5 enterprise accounts ## Audit (Structural Integrity Auditor) ⚠ Warning: No pricing model specified. What's the flat fee? ⚠ Warning: "40% higher retention" — do you have a baseline? ✔ Hypothesis is testable with a small pilot ✔ Clear user segment defined Saved to vault/Hypotheses/hypothesis-enterprise-subscription.md Updated context index.
What just happened
  1. Claude read your strategic bets and connected the idea to your existing context
  2. It framed a proper hypothesis using the "We believe... for... will... because" format
  3. It extracted hidden assumptions and ranked them by risk
  4. It defined success criteria so you know how to test it
  5. The Structural Integrity Auditor automatically checked for gaps
  6. The hypothesis itself is saved to your vault for future reference

Go check Obsidian — open Hypotheses/ and you'll see the new file.

Exercise

Build Your Own Hypothesis

Think of a real idea for your product. Run /hypothesis-builder and describe it to Claude. See how it frames it, extracts assumptions, and connects it to your strategy.

7

Run Your First Playbook

Playbooks are the real power of AI-PHISR. They chain multiple skills into a multi-step workflow. One command, many outputs.

Let's Prep for a 1:1

This is one of the most immediately useful playbooks. It reads your initiatives, open decisions, stakeholder info, sprint state, and recent context, then builds a decision-focused agenda.

Type this at the Claude Code prompt /playbook-manager-1on1
Best results once your vault has some context This playbook works best after you have active initiatives, at least one open decision, your manager details in the worksheet, and a current sprint state. If some of that is missing, Claude may ask follow-up questions or produce a thinner agenda.
Terminal — Claude Code
You: /playbook-manager-1on1 Step 1/5: Loading initiatives and sprint state... Step 2/5: Scanning for open decisions... Step 3/5: Checking for blockers... Step 4/5: Building decision-focused agenda... Step 5/5: Preparing talking points... ## 1:1 Agenda — Meeting with [Your Manager] Estimated time: 25 minutes 1. Decision Needed: Subscription pricing model (10 min) Context: Enterprise customers requesting subscriptions. We have a hypothesis but no pricing model yet. Your recommendation: Start with $15/employee/month pilot. Ask: Do you agree with piloting at this price point? 2. Escalation: Kitchen partner onboarding bottleneck (5 min) Context: Only 8 of target 15 kitchen partners signed. Blocker: Legal review taking 2x longer than expected. Ask: Can you help expedite legal review? 3. Strategic Question: Q2 bet allocation (8 min) Context: Micro-kitchen bet showing early promise (12 min avg delivery). Curated menu less clear. Question: Should we double down on kitchens in Q2? 4. Quick Highlights (2 min) - Delivery times down to 12 min avg (from 18 min) - Enterprise pilot: 3 companies confirmed
What just happened Claude read the relevant vault context — initiatives, decisions, sprint state, stakeholder info, and recent context — and built a decision-first agenda. This playbook is primarily an output workflow, not an automatic save-to-vault workflow.

Try Another Playbook

Here's the full list. Try whichever one matches your current situation:

Command When to Use
/playbook-new-ideaSomeone has a new request or feature idea
/playbook-sprint-kickoffStarting a new sprint
/playbook-launchAbout to ship something, need go/no-go
/playbook-incidentSomething broke in production
/playbook-manager-1on1Before your 1:1 with your manager
/playbook-weekly-reviewEnd of the week
/playbook-pitchNeed to convince someone of something
/playbook-post-launch-dataAnalyzing results after a launch
About sprint kickoff /playbook-sprint-kickoff works best after your strategic bets, OKRs, initiatives, sprint state, and dependencies are in place. If that context is still thin, expect follow-up questions or a lighter output on the first run.
Exercise

Run a Playbook That Fits Your Week

Pick the playbook that matches something you actually need to do this week. Run it and see how AI-PHISR uses your product context to generate useful, specific output.

8

Your Daily Workflow

Now that you have a working system, here's how to use it every day. The key insight: the more you use it, the smarter it gets, because the most important outputs get saved to your vault and the memory files keep useful continuity.

Morning Routine (2 minutes)

Open Claude Code and ask about your day:

Terminal — Claude Code
$ claude You: What should I focus on today? Any blockers or urgent decisions? Today's Priorities (Sprint day 6 of 10): ⚠ Needs your attention: 1. Subscription pricing decision — still open, your manager expects a proposal by Thursday 2. Kitchen partner #9 contract — legal review stuck ★ In progress: - Menu curation algorithm: on track (day 3 of 5) - Enterprise pilot outreach: 3 of 5 confirmed 📅 Today's meetings: - 10:00 AM: 1:1 with manager (run /playbook-manager-1on1 to prep) - 2:00 PM: Kitchen partner review

When Something Happens During the Day

New idea comes in

Type: /playbook-new-idea

Need to make a decision

Type: /tradeoff-mapper

Prepping for a meeting

Type: /1on1-prep

Something broke

Type: /playbook-incident

End of week

Type: /playbook-weekly-review

Need to pitch something

Type: /playbook-pitch

The Compounding Effect

Here's what happens over time:

Week 1
Claude knows your strategy, bets, and team
Week 2
+ It knows your sprint history and decisions you've made
Week 3
+ It has validated/invalidated hypotheses and research insights
Month 2+
It knows your product better than any colleague. Context never resets.

Quick Tips

Final Project

Build Your Complete Product System

By now you have the tools and knowledge. Complete these tasks to make AI-PHISR fully operational for your product:

  • Fill out Setup_Worksheet.md completely (all sections)
  • Run /populate-strategy to bootstrap the vault
  • Optionally run /pm-voice-creator after you have real writing samples
  • Create at least 2 initiative files with full details
  • Build 1 hypothesis using /hypothesis-builder
  • Log 1 open decision in vault/Decisions/
  • Run /playbook-manager-1on1 before your next 1:1
  • Run /playbook-weekly-review at the end of the week
  • Ask Claude "what should I focus on today?" on 3 different days

After completing these, you'll have a fully operational PM system that remembers everything and gets smarter every day.

Congratulations! You've completed the AI-PHISR course. You now have a PM operating system that maintains persistent context about your product. Every decision you log, every hypothesis you test, every sprint you summarize makes it more powerful. Welcome to compounding AI intelligence.