whatsmyedge/Kenji Watanabe — product marketing manager at a B2B SaaS startup, 34, tech-adjacent, EN intake
Model claude-sonnet-4-6Elapsed 9m 24sTokens 14.3k↑ 32.1k↓tool_use
Your Protocol

Kenji Watanabe

Product Marketing Manager, B2B SaaS (Devtools)
B2B SaaS Product Marketing — Developer Tools
Your mission: become the PMM that your organization cannot afford to lose and the market cannot afford to ignore — by proving that AI systems under your direction generate more revenue, more pipeline, and more competitive clarity than any tool or hire could replace you with.
Plan ID
PR-2026-0847
Tier
PREMIUM
Duration
30 days
Daily
90 min
Level
2 → 3
Generated
2026-04-22
Executive summary

The 30-day bet in one page

Kenji, here is the strategic situation as this Protocol reads it.

Situation Analysis

You are a solo PMM at a Series-B devtools startup in a market where AI is compressing the execution layer of product marketing faster than most practitioners realize. The tasks that historically justified a PMM headcount — writing launch posts, drafting battlecards, monitoring competitors, producing enablement decks — are now tasks that AI can do in minutes with the right system behind them. You've already internalized this. Your Zapier pipeline, your Custom GPT, your Retool dashboard: these are not experiments. They are early evidence that you understand the shift.

The risk is not that you'll be replaced. Your vulnerability score of 36 is below the threshold where replacement is the primary threat. The risk is subtler: that your AI-built systems remain invisible to leadership, that your newsletter stays at 400 subscribers while someone less capable builds the audience you should own, and that your two gaps — building as code and presenting to leadership — leave you unable to scale the systems or claim credit for them.

Recommended Focus Areas

First: the competitive intelligence system. You've described building an AI-powered CI system for the sales team as one of your three strategic priorities. This is the highest-leverage project in your portfolio. A well-built CI system that sales reps actually use is a revenue artifact — it shortens deal cycles, improves win rates, and makes you the person who built the thing that closes deals. This Protocol dedicates Days 8-18 to building it at production quality, with memory, self-evaluation, and a financial model that quantifies its impact.

Second: Operators Playbook monetization. You have 400 subscribers and a clear thesis. The gap between a newsletter and a paid community is not content — it's architecture. This Protocol builds the monetization architecture in Days 22-28: a paid tier structure, a founding member offer, a course outline, and a launch sequence. The newsletter becomes the proof-of-work that makes the Leap credible.

Third: leadership presentation. Your gap here is not about slide design. It's about translating AI system outputs into business outcomes that a VP or CFO cares about. Days 19-21 are dedicated to building the business case and practicing the presentation with AI as your sparring partner.

Key Risks

The data constraint around Gong and Amplitude is real and must be respected throughout. Every system built in this Protocol uses public-source or synthetic data. The CI system is built on public competitor signals — G2, pricing pages, job postings, release notes — not on internal call data. This is actually a feature: a CI system that runs on public signals is one you can describe publicly without NDAs.

The second risk is scope creep. You have three strategic priorities, a newsletter, a pricing overhaul, and a 30-day protocol. The Protocol sequences these deliberately — do not run them in parallel. The pricing overhaul is referenced as context but is not a Protocol deliverable; it would dilute the arc.

Opportunity Map

The devtools AI space is moving fast enough that a PMM who can credibly write about AI-native go-to-market strategy — and demonstrate it with working systems — has a narrow but real window to become the definitive voice in that niche. Operators Playbook is positioned correctly. The opportunity is to close the gap between "newsletter writer" and "practitioner who ships systems and teaches others how." This Protocol is designed to close that gap in 30 days.

Priority matrix

What to fight for first

A 2×2 of urgency × impact. Do-first items are irreversible; Park items are noise in disguise.

Do first

High impact · High urgency
  • AI-powered competitive intelligence system (high revenue impact, systems already partially built)
  • Leadership business case for AI PMM investment (high visibility, leverages existing work)
  • Operators Playbook paid tier architecture (high income potential, audience already exists)
  • Prompt library productization — convert personal library into team-usable system

Schedule

High impact · Lower urgency
  • Code-level automation: Python scripts for CI data ingestion (high impact, requires skill-building)
  • Operators Playbook course curriculum design (high revenue ceiling, significant build time)
  • Multi-source CI system with memory and self-evaluation (high impact, 2-3 week build)
  • Sales enablement system with AI-generated battlecard refresh loop (high sales impact, complex integration)

Quick wins

Low effort · Fast payoff
  • Audit and sharpen existing 20-prompt library — remove weak prompts, document the strong ones
  • Set up Perplexity Pro saved searches for top 5 competitors — 15 minutes, daily signal
  • Draft Operators Playbook founding member offer page using Claude Pro
  • Run The Method on one existing battlecard — validate, iterate, improve in one session

Park

Low impact · Low urgency
  • Fine-tuning a custom model on company messaging (high effort, marginal gain over prompt engineering)
  • Building a full Retool app for the CI system (Notion-based version ships faster and is more shareable)
  • Redesigning the newsletter template from scratch (content and monetization matter more than design now)
  • Automated social posting pipeline (premature — positioning strategy comes first)
Profile

Where you stand today

Kenji, you're a solo Product Marketing Manager at an 80-person Series-B devtools startup — the person who owns positioning, competitive intelligence, launch narratives, and sales enablement all at once, with no team beneath you. You report to a VP Marketing, work cross-functionally with product, sales, and CS, and you've already built more with AI than most PMMs at companies three times your size.

You're not here because you're behind. You're here because you can see the gap between where you are and where you could be — and it's making you impatient.

Your AI usage is already at the "full workflows" level. You've built a Custom GPT for battlecard drafting, a Zapier pipeline that surfaces G2 mentions daily, a Retool dashboard that flags churn-risk accounts using Amplitude and the OpenAI API, and a Claude prompt library of roughly 20 reusable launch messaging frameworks. That's a Level 2 operator. The gap to Level 3 isn't about learning more tools — it's about shifting from personal productivity to systems that generate leverage at organizational scale, and then packaging that leverage into something the market will pay for.

Your two goals — Build and Leap — are not in tension. They're the same move. The newsletter you're building ("Operators Playbook," 400 subscribers, covering AI for PMMs and product people) is the proof-of-work artifact that makes the Leap credible. The systems you build inside your company are the case studies that make the newsletter worth paying for. This Protocol treats both goals as one integrated arc.

Your tool stack is strong: ChatGPT Enterprise and Notion AI for company work, Claude Pro and Perplexity Pro personally, Cursor for Retool tinkering, Zapier and Make for automation. Your constraints are real: customer interview recordings stay in Gong, product data stays in Amplitude and Snowflake. This Protocol respects those boundaries — every data task uses synthetic or anonymized examples, and your competitive intelligence systems are built on public-source inputs only.

Your two declared gaps — building things as code and presenting to leadership — are the exact two skills that separate a PMM who is valuable from a PMM who is irreplaceable. Building as code means your systems survive your absence and scale beyond what a prompt library can do. Presenting to leadership means your work gets resourced, protected, and attributed. Both gaps get direct, daily attention here.

Your vulnerability score of 36 reflects a real but manageable exposure. The tasks most at risk — writing launch posts, drafting battlecards from public sources, monitoring competitor changes — are exactly the tasks you've already started automating. The strategic work you're doing (pricing overhaul, AI-powered competitive intelligence system, thought leadership content program) is where your irreplaceability lives. This Protocol accelerates that transition: from executing tasks to designing the systems that execute them, and then making that capability visible to the market.

Thirty days. Ninety minutes a day. By the end, you'll have a working competitive intelligence system, a monetization architecture for Operators Playbook, a leadership-ready business case, and a public positioning that makes you the obvious person to hire, follow, or partner with in the devtools AI space."

Level assessment

Your current level & trajectory

L2
AI-Integrated
You use AI daily across your workflow and have built functional systems — a Custom GPT, a Zapier pipeline, a Retool dashboard, a reusable prompt library. You're generating real output faster than most peers. The limitation at Level 2 is that your systems are personal: they live in your head, your accounts, and your Notion. They don't scale to the team, don't generate revenue independently, and don't make you visible to the market. Level 3 is the shift from personal productivity to organizational leverage — designing systems others depend on, packaging your methods into products, and building a public track record that compounds.
Days 1-7
Compress and sharpen your existing AI stack. Audit what you've built, find the gaps, and rebuild your top systems with the rigor of someone who will hand them to a team — or sell them to a market.
Days 8-21
Build the AI-powered competitive intelligence system at production quality. Add memory, self-evaluation, and a financial model. Begin translating internal systems into external positioning.
Days 22-30
Ship the Operators Playbook monetization architecture. Build the leadership business case. Publish the positioning that makes you the go-to voice for AI-native PMMs in the devtools space.
Exposure diagnosis

Where AI meets your work

Every task scored across five dimensions. Higher numbers mean AI can substitute for more of the task today; lower numbers mean your judgment is still load-bearing.

14Tasks scored
5 exposed. 6 contested. 3 safe.Summary
Your exposure map shows a PMM whose execution layer is heavily automatable but whose strategic layer is genuinely defensible. The five exposed tasks — writing launch posts, drafting battlecards from public sources, monitoring competitor pricing and features, producing enablement decks, and summarizing customer interview recordings — are exactly the tasks you've already started automating. This is not a coincidence: you built those systems because you felt the exposure. The contested tasks are where the real leverage lives: the systems you've built are good, but they're not yet production-quality or team-scale. The three safe tasks reflect your irreplaceable judgment layer. These scores reflect AI capabilities as of April 2026. The trajectory is consistently toward greater automation — tasks scored as contested today may cross into exposed within 6-12 months. The plan ahead accelerates your move into the safe zone before that happens.
The Protocol sequences your work to match this map: Phase 1 hardens your automation of exposed tasks, Phase 2 builds production-quality systems for contested tasks, and Phase 3 makes your safe-zone capabilities visible to leadership and the market.
Writing launch blog posts and one-pagersAI-Leveraged
High pattern-match to training data; AI produces credible first drafts that require only light editing for brand voice.
Pattern → Judgment88
Processing → Understanding80
Analysis → Gut Feel75
Execution → Decision85
Output → Ownership82
82
Exposure
Drafting competitive battlecards from public sourcesAI-Leveraged
Public-source synthesis is a core AI strength; the judgment layer (what to emphasize for sales) is thin and learnable.
Pattern → Judgment82
Processing → Understanding85
Analysis → Gut Feel70
Execution → Decision78
Output → Ownership80
79
Exposure
Monitoring competitor pricing and feature changes weeklyAI-Leveraged
Pure information processing and pattern detection — AI does this faster, more consistently, and at greater scale than any human.
Pattern → Judgment92
Processing → Understanding90
Analysis → Gut Feel80
Execution → Decision88
Output → Ownership90
88
Exposure
Producing sales enablement decks for new feature launchesAI-Leveraged
Structure and content are highly templatable; AI handles the heavy lifting, leaving only strategic framing and polish.
Pattern → Judgment78
Processing → Understanding72
Analysis → Gut Feel68
Execution → Decision76
Output → Ownership75
74
Exposure
Summarizing customer interview recordings into messaging insightsAI-Leveraged
Gong's AI summaries already do this; the synthesis step is the only remaining human layer, and it's thin.
Pattern → Judgment75
Processing → Understanding78
Analysis → Gut Feel65
Execution → Decision70
Output → Ownership68
71
Exposure
Managing product roadmap communications to sales and CSCo-pilot Territory
The writing is automatable but the judgment about what to emphasize, what to withhold, and how to sequence for each audience requires organizational context AI doesn't have.
Pattern → Judgment62
Processing → Understanding55
Analysis → Gut Feel50
Execution → Decision60
Output → Ownership62
58
Exposure
Designing pricing and packaging strategyCo-pilot Territory
AI can model scenarios and surface frameworks, but the judgment about market positioning, competitive response, and internal politics is deeply human.
Pattern → Judgment48
Processing → Understanding45
Analysis → Gut Feel35
Execution → Decision40
Output → Ownership42
42
Exposure
Building AI-powered competitive intelligence systemCo-pilot Territory
The system design, source selection, and signal interpretation require PMM judgment; the data collection and synthesis are automatable.
Pattern → Judgment42
Processing → Understanding40
Analysis → Gut Feel32
Execution → Decision38
Output → Ownership38
38
Exposure
Developing thought leadership content programCo-pilot Territory
Content drafting is automatable; the editorial judgment, audience understanding, and original thesis generation are not.
Pattern → Judgment50
Processing → Understanding48
Analysis → Gut Feel38
Execution → Decision44
Output → Ownership45
45
Exposure
Cross-functional stakeholder alignment (product, sales, CS)Co-pilot Territory
AI can draft communications and prep materials, but the relationship navigation, trust-building, and real-time negotiation are human.
Pattern → Judgment38
Processing → Understanding32
Analysis → Gut Feel30
Execution → Decision36
Output → Ownership38
35
Exposure
Operators Playbook newsletter editorial strategyCo-pilot Territory
Writing is AI-assisted but the thesis selection, audience intuition, and voice are yours — and they're the product.
Pattern → Judgment45
Processing → Understanding42
Analysis → Gut Feel35
Execution → Decision38
Output → Ownership40
40
Exposure
Positioning and messaging strategy for new product areasHuman-Judgment Core
Requires deep market intuition, customer empathy, and competitive foresight that AI can support but not replace.
Pattern → Judgment25
Processing → Understanding22
Analysis → Gut Feel18
Execution → Decision20
Output → Ownership25
22
Exposure
Presenting AI investment cases to VP and leadershipHuman-Judgment Core
Organizational credibility, political navigation, and real-time persuasion are irreducibly human — AI can prep you but cannot replace you in the room.
Pattern → Judgment20
Processing → Understanding18
Analysis → Gut Feel15
Execution → Decision18
Output → Ownership20
18
Exposure
Operators Playbook community and course product designHuman-Judgment Core
The product vision, community design, and curriculum judgment are yours — they're what the market is paying for.
Pattern → Judgment18
Processing → Understanding15
Analysis → Gut Feel12
Execution → Decision15
Output → Ownership15
15
Exposure
Phase 0

Replaceability Diagnosis

Replacement risks

  • Launch blog posts and one-pagers: a well-prompted GPT-5 or Claude Opus 4.7 with your brand voice guidelines can produce a publish-ready first draft in under 3 minutes. If this is where your time goes, it's the first thing a cost-cutting conversation will point to.
  • Competitive battlecard drafting: your Custom GPT is already doing this — which means the question is no longer 'can AI do this?' but 'does the company still need you to supervise it?' The answer is yes, but only if you're the one who designed the system and interprets its outputs.
  • Competitor monitoring: the Zapier pipeline you built for G2 mentions is a proof-of-concept. A more sophisticated system — one that ingests pricing pages, job postings, release notes, and review sites simultaneously — could run entirely without you. Someone will build it. It should be you.
  • Sales enablement deck production: templated decks for feature launches are increasingly automatable end-to-end. If your value is in producing the deck rather than designing the narrative strategy behind it, that value is compressible.
  • Customer interview summarization: Gong's AI already does this. The risk is that leadership starts to wonder what the PMM layer adds if the synthesis is automated — unless you're the one who designed the synthesis framework and acts on the insights.

Irreplaceable traits

  • Positioning judgment: you know which message will land with a developer audience versus a VP of Engineering versus a CFO — and you know when the same product needs three different stories. AI can generate variants; only you can pick the right one.
  • Organizational context: you know what the sales team is actually struggling with, what the CS team is hearing from churned accounts, and what the product roadmap will and won't deliver. AI has none of this without you feeding it — and feeding it well is a skill.
  • System design: you've already built four AI systems. The ability to design, build, and iterate on AI-powered workflows is a skill that compounds. Most PMMs can use AI tools; you can build the tools that other PMMs use.
  • Operators Playbook: 400 subscribers who trust your editorial judgment on AI and product marketing. That audience is yours. No AI system can build that relationship for you — it can only help you serve it better.
  • Cross-functional credibility: you sit at the intersection of product, sales, CS, and marketing. The trust you've built in those relationships — knowing when to push, when to defer, when to reframe — is organizational capital that takes years to accumulate and cannot be automated.

The five exposed tasks on your map are not your identity — they're your overhead. This Protocol's first move is to harden the automation of those tasks so completely that they stop consuming your attention. The contested tasks are where you'll build the systems that make you organizationally essential. The safe tasks are where you'll build the public positioning that makes you market-valuable. By Day 30, the exposed tasks will be running on systems you designed, the contested tasks will be supported by AI infrastructure you built, and the safe tasks will be visible to the market through Operators Playbook and your leadership track record. That's the arc.

Daily protocol

Your 30-day plan, day by day

Each day: one task, one artifact, one prompt. Click any day to expand. Day 1 and milestones open by default.

1Day
The Method + Your AI Stack Audit
Phase 1Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Produce a written audit of your existing AI systems and a printed Method Reference Card you'll use for the next 30 days.
  1. Print or save the Method Reference Card artifact below — this is your operating system for every day ahead.
  2. Open Claude Pro and paste the prompt template to run a structured audit of your four existing AI systems.
  3. For each system, complete the 5-step loop: Describe what it does, Generate the audit output, Validate against your real experience, Iterate on any gaps, Review the full picture.
  4. Save the audit as a Notion page titled 'AI Stack Audit — Day 1'.
Starter prompt
I am a Product Marketing Manager at a B2B SaaS devtools startup. I have built four AI systems: (1) a Custom GPT for competitor battlecard drafting, (2) a Zapier workflow that summarizes daily G2 review mentions, (3) a Retool dashboard that surfaces churn-risk accounts using product analytics and the OpenAI API, and (4) a Claude prompt library of roughly 20 reusable launch messaging frameworks.

For each system, give me a structured audit with these five fields:
- WHAT IT DOES: one sentence
- WHAT IT DOES WELL: two bullet points
- WHERE IT BREAKS: two bullet points (be specific about failure modes)
- WHAT IT IS MISSING: one gap that would make it 10x more useful
- UPGRADE PRIORITY: High / Medium / Low with one sentence of reasoning

After the four audits, give me a one-paragraph summary of which system to upgrade first and why, from the perspective of a PMM who wants to become irreplaceable to their organization.
Artifact
Day1-AI-Stack-Audit.notion doc
You have a Notion page with four structured audits and a prioritized upgrade recommendation. You've disagreed with at least one of the AI's assessments and written your correction in the doc.
Reflection — 3 questions
  1. Which of your four AI systems did the audit say was weakest — and do you agree? What did you override?
  2. What surprised you most about how the AI described your existing systems?
  3. Which system, if it broke tomorrow, would cause the most pain? Is that the one you're planning to upgrade first?
You can't build forward without an honest map of where you are — this audit is the foundation every subsequent day builds on.
2Day
Stress-Test Your Best System
Phase 1Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Uncomfortable
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Find the failure modes in your Custom GPT battlecard system by deliberately feeding it bad, incomplete, and adversarial inputs.
  1. Open your Custom GPT for competitor battlecard drafting.
  2. Run three stress-test scenarios: (a) a competitor with almost no public information, (b) a competitor who just pivoted and whose old positioning is now wrong, (c) a fictional competitor you invent with contradictory signals.
  3. For each scenario, document: what the GPT produced, where it hallucinated or failed, and what a human would have caught.
  4. Paste your findings into Claude Pro and ask it to generate a 'Failure Mode Registry' — a structured list of conditions under which the system breaks.
  5. Save the registry as a Notion page. This becomes your validation checklist for every battlecard the system produces.
Starter prompt
I have an AI system (a Custom GPT) that drafts competitive battlecards for a B2B SaaS devtools product. I ran three stress tests on it today and documented the following failure cases:

[PASTE YOUR STRESS TEST FINDINGS HERE — what the GPT produced, what was wrong, what a human would have caught]

Based on these failure cases, generate a structured Failure Mode Registry with the following format for each failure mode:
- TRIGGER CONDITION: what input or situation causes this failure
- FAILURE TYPE: hallucination / stale data / missing context / wrong framing / other
- DETECTION METHOD: how a human reviewer would catch this
- MITIGATION: one specific change to the prompt or process that would reduce this failure

End with a one-paragraph 'Validation Protocol' — a checklist a sales rep or PMM should run on every battlecard before using it in a deal.
Artifact
Day2-Battlecard-Failure-Mode-Registry.notion checklist
You have a Failure Mode Registry with at least 5 distinct failure modes, each with a detection method and mitigation. The validation protocol is specific enough that a sales rep could use it without your help.
Reflection — 3 questions
  1. What was the most embarrassing failure the stress test revealed — something that could have hurt a deal if it slipped through?
  2. Which failure mode was hardest to detect without domain expertise? What does that tell you about the irreplaceable layer in your work?
  3. What you're building could change how work gets done around you. That's real. Sit with it for 30 seconds. Then keep going — because someone will build this whether you do or not, and it's better if it's someone who understands the human side. What's one thing you'll do to make sure the human judgment layer stays visible in the systems you build?
A system you can't stress-test is a system you can't trust — and a system you can't trust will eventually embarrass you in front of a customer or a VP.
3Day
Rebuild the Battlecard System with Memory
Phase 1Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Redesign your competitor battlecard Custom GPT with a structured memory layer so it retains competitive context across sessions.
  1. Open Claude Pro and use the prompt template to design a memory architecture for your battlecard system.
  2. Create a 'Competitor Context Document' in Notion — a structured template that stores: competitor name, last-updated date, pricing tier, top 3 differentiators, top 3 weaknesses, recent moves (last 90 days), and win/loss notes.
  3. Fill in the template for your top 3 competitors using only public sources (pricing pages, G2, release notes, job postings).
  4. Update your Custom GPT system prompt to reference this Notion document as its 'source of truth' — paste the current context into each session as a preamble.
  5. Run a test battlecard generation using the new memory-augmented prompt. Compare output quality to Day 1 baseline.
Starter prompt
I am rebuilding an AI-powered competitive battlecard system for a B2B SaaS devtools product. The current system has no memory — each session starts from scratch. I want to add a structured memory layer.

Design a 'Competitor Context Document' template I can maintain in Notion that will serve as the system's persistent memory. The template should:
- Be fillable from public sources only (pricing pages, G2 reviews, job postings, release notes, LinkedIn)
- Include fields that are most predictive of win/loss outcomes in B2B SaaS sales
- Be updatable in under 20 minutes per competitor per week
- Be structured so I can paste it directly into a GPT session as a preamble

Also write a system prompt preamble (under 300 words) that instructs the GPT how to use the context document when generating battlecards. The preamble should tell the GPT to flag when context is stale (older than 30 days) and ask for an update before generating.
Artifact
Day3-Competitor-Context-Template.notion doc
You have a filled Competitor Context Document for 3 competitors and a revised Custom GPT system prompt. Running the new system produces a battlecard that references specific, dated competitor moves — not generic claims.
Reflection — 3 questions
  1. How different was the battlecard quality with the memory layer versus without it? What specifically changed?
  2. Which field in the Competitor Context Document was hardest to fill from public sources — and what does that gap tell you about your competitive intelligence blind spots?
  3. If you handed this system to a new sales hire tomorrow, what would they get wrong? What's still in your head that isn't in the system?
A battlecard system with memory is a system that gets smarter over time — and one that a team can use without you in the room.
4Day
Build the Public-Signal CI Pipeline
Phase 1Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Design and document a multi-source competitive intelligence pipeline that ingests public signals daily and surfaces them in a structured format for sales.
  1. Use Claude Pro to design the architecture of a CI pipeline that ingests: G2 reviews, competitor pricing pages, competitor job postings, competitor release notes/changelogs, and LinkedIn posts from competitor executives.
  2. Map each source to a Zapier or Make trigger you already have access to, or identify which ones need to be added.
  3. Create a 'CI Signal Taxonomy' — a classification system for incoming signals (Pricing Change / Feature Launch / Hiring Signal / Customer Sentiment / Executive Messaging).
  4. Draft a Notion database schema for storing and tagging incoming signals.
  5. Write the 'CI Digest' template — the weekly summary format that gets sent to sales and CS.
Starter prompt
I am a Product Marketing Manager building an AI-powered competitive intelligence system for a B2B SaaS devtools startup. The system should run on public signals only (no internal data). My automation stack includes Zapier and Make.

Design the architecture for a multi-source CI pipeline with these requirements:
- Sources: G2 reviews, competitor pricing pages, competitor job postings, competitor release notes or changelogs, LinkedIn posts from competitor executives
- Output: a structured weekly digest for sales and CS teams
- Automation: identify which Zapier or Make triggers would handle each source
- Classification: a signal taxonomy that categorizes each incoming item

Deliver:
1. A pipeline architecture diagram described in plain text (boxes and arrows, no code)
2. A CI Signal Taxonomy with 5-7 categories and one example per category
3. A Notion database schema (field names, field types, and purpose of each field)
4. A weekly CI Digest template (the format sales reps will actually read)

Keep the design opinionated — tell me which sources matter most for a devtools competitive landscape and why.
Artifact
Day4-CI-Pipeline-Architecture.notion doc
You have a pipeline architecture, a signal taxonomy, a Notion schema, and a CI Digest template — all specific enough that you could hand them to a contractor and have them build it without a briefing call.
Reflection — 3 questions
  1. Which signal source will be hardest to automate reliably — and what's your manual fallback for that one?
  2. How did the AI's architecture differ from what you would have designed yourself? What did you keep, what did you change?
  3. Who in sales or CS would benefit most from this digest? Have you told them it's coming?
Compliance note
This pipeline uses public-source data only. Do not connect it to Gong, Amplitude, or Snowflake — those data sources require separate compliance review.
A CI system that runs on public signals is one you can describe publicly, demo to leadership, and hand to a team — it's the foundation of your organizational leverage.
5Day
Productize Your Prompt Library
Phase 1Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Transform your 20-prompt Claude library from a personal tool into a documented, team-usable system with a quality rating and usage guide.
  1. Open your existing prompt library and run each prompt through a rapid quality assessment: Does it produce consistent output? Does it require domain knowledge to use correctly? Is it documented well enough for someone else to use?
  2. Use Claude Pro to generate a structured 'Prompt Library Audit' — rate each prompt on consistency, transferability, and documentation quality.
  3. Identify your top 5 highest-value prompts and write a one-paragraph usage guide for each: when to use it, what inputs it needs, what the output looks like, and what to watch out for.
  4. Create a Notion page titled 'PMM Prompt Library v2' with the top 5 documented prompts, the full list with ratings, and a 'How to Use This Library' intro section.
  5. Tag each prompt by use case: Positioning / Competitive / Launch / Enablement / Research.
Starter prompt
I have a library of approximately 20 reusable prompts I've built for product marketing tasks including launch messaging, competitive positioning, sales enablement, and customer research synthesis. I want to turn this into a documented, team-usable system.

Here are my top prompts (paste your prompts here, one at a time or in a batch):
[PASTE YOUR PROMPTS HERE]

For each prompt, give me:
1. QUALITY RATING: Consistency (1-5), Transferability (1-5), Documentation (1-5)
2. USE CASE TAG: Positioning / Competitive / Launch / Enablement / Research
3. USAGE GUIDE: When to use it, what inputs it needs, what the output looks like, one thing to watch out for
4. IMPROVEMENT: One specific change that would make this prompt 20% more reliable

After all prompts, give me a 'How to Use This Library' intro section (under 200 words) written for a PMM who didn't build these prompts and is using them for the first time.
Artifact
Day5-PMM-Prompt-Library-v2.notion doc
The library has a 'How to Use This Library' intro, at least 5 prompts with full usage guides, and every prompt has a quality rating and use case tag. A new PMM hire could use it on Day 1 without asking you for help.
Reflection — 3 questions
  1. Which prompt in your library is the most powerful but the hardest to transfer to someone else — and why?
  2. What's missing from your library that you find yourself rebuilding from scratch every few weeks?
  3. If this library were a product, what would you charge for it? Who would buy it?
A prompt library that only you can use is a personal tool; one that anyone on the team can use is organizational infrastructure — and infrastructure is what makes you irreplaceable.
6Day
Automate the Launch Enablement Deck
Phase 1Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Build a reusable AI-powered workflow that produces a first-draft sales enablement deck for any new feature launch in under 15 minutes.
  1. Use Claude Pro to design a 'Launch Enablement Deck' prompt template — one that takes a feature brief as input and outputs a structured deck outline with slide titles, talking points, and objection-handling notes.
  2. Test the template against a real or recent feature launch. Run the full 5-step loop: Describe the feature, Generate the deck outline, Validate against what sales actually needs, Iterate on the objection-handling section, Review the full deck.
  3. Create a Google Slides or Notion template that maps to the AI-generated outline — so the output drops directly into a usable format.
  4. Document the end-to-end workflow: what inputs the PMM provides, what the AI produces, what the PMM edits, and how long each step takes.
  5. Save the workflow documentation and the deck template as paired artifacts.
Starter prompt
I am a Product Marketing Manager at a B2B SaaS devtools startup. I need to create a reusable prompt template that generates a first-draft sales enablement deck outline for any new feature launch.

The deck is used by account executives in discovery and demo calls with technical buyers (developers, engineering managers, CTOs at SMB and mid-market companies).

Here is the feature brief for the launch I want to test this on:
[PASTE FEATURE NAME, ONE-PARAGRAPH DESCRIPTION, AND TOP 3 CUSTOMER PROBLEMS IT SOLVES]

Generate a sales enablement deck outline with:
- Slide 1: The Problem (what the customer is experiencing today)
- Slide 2: Why Now (what changed in the market or in our product that makes this the right time)
- Slide 3: The Solution (what the feature does, in language a developer will trust)
- Slide 4: How It Works (3-step process or architecture diagram description)
- Slide 5: Proof Points (what to say when asked 'does this actually work?')
- Slide 6: Objection Handling (top 3 objections and responses)
- Slide 7: Call to Action (what the AE asks for at the end of this conversation)

For each slide, give me: slide title, 3-5 bullet talking points, and one 'do not say' warning.

After the outline, give me the reusable prompt template (with [bracket] placeholders) that I can use for any future feature launch.
Artifact
Day6-Launch-Enablement-Deck-Workflow.notion doc
You have a prompt template that produces a structured 7-slide deck outline from a feature brief, a documented workflow, and a Notion or Slides template that the output maps to. Time from feature brief to deck outline is under 15 minutes.
Reflection — 3 questions
  1. Which slide did the AI get most wrong on the first pass — and what did you have to add from your own knowledge of the sales team?
  2. How does this workflow change what you'd spend your time on during a launch week?
  3. If you ran this workflow for every feature launch for the next quarter, what would you do with the time you saved?
A repeatable launch workflow means you can support more launches, more consistently, without adding headcount — and that's a business case you can take to your VP.
7Day
Week 1 Milestone — The Method in Review
Phase 1Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Milestone
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Produce a Week 1 retrospective document that maps everything you've built to The Method and identifies the highest-leverage upgrade for Week 2.
  1. Open Claude Pro and paste the prompt template to generate a structured retrospective of your Week 1 artifacts.
  2. For each of the six artifacts you built (Stack Audit, Failure Mode Registry, Competitor Context Template, CI Pipeline Architecture, Prompt Library v2, Launch Enablement Workflow), assess: which step of The Method was hardest, and what you'd do differently.
  3. Identify the single highest-leverage upgrade for Week 2 — the one system that, if built properly, would have the most impact on your organization and your positioning.
  4. Write a 'Week 2 Intention' — one paragraph stating what you're building, why it matters, and what done looks like.
  5. Save the retrospective as your milestone document.
Starter prompt
I am a Product Marketing Manager who has just completed the first week of an AI Irreplaceability Protocol. Here is a summary of what I built this week:

1. AI Stack Audit — reviewed my four existing AI systems and identified upgrade priorities
2. Battlecard Failure Mode Registry — stress-tested my Custom GPT and documented where it breaks
3. Competitor Context Template — added a memory layer to my battlecard system
4. CI Pipeline Architecture — designed a multi-source competitive intelligence pipeline
5. PMM Prompt Library v2 — documented and rated my 20 reusable prompts
6. Launch Enablement Deck Workflow — built a reusable workflow for feature launch decks

For each artifact, I want you to:
- Identify which step of the 5-step Method (Describe / Generate / Validate / Iterate / Review) was the hardest based on the type of work involved
- Give me one specific upgrade that would make this artifact 2x more useful in Week 2

Then give me:
- A ranking of the six artifacts by organizational impact (most to least)
- A recommendation for the single highest-leverage system to build in Week 2, with a one-paragraph rationale
- Three questions I should be able to answer by the end of Week 2 to know I'm on track
Artifact
Day7-Week1-Retrospective-Milestone.notion doc
You have a retrospective that names all six artifacts, rates them by organizational impact, and commits to a specific Week 2 build target. The Week 2 Intention paragraph is specific enough that you could share it with your VP as a preview.
Reflection — 3 questions
  1. Which of the six artifacts from Week 1 are you most proud of — and which one are you least confident in?
  2. By the time you reach Day 30, the AI landscape will have shifted again — but you'll have The Method to adapt. What's one thing you learned this week that you'll still be using in 6 months, regardless of which tools exist?
  3. What's the gap between what you built this week and what you'd need to show your VP to get a 'yes' on resourcing the CI system?
Week 1 built the foundation; Week 2 builds the system that makes you organizationally essential.
Milestone

Week 1 Complete: From Stack to System

  • Audited and stress-tested all four existing AI systems — found and documented real failure modes
  • Added a memory layer to the battlecard Custom GPT using a structured Competitor Context Document
  • Designed a multi-source CI pipeline architecture ready for implementation in Week 2
  • Productized the 20-prompt library into a team-usable, documented system
Solidly at Level 2 — personal AI systems are hardened and documented. Beginning transition to Level 3 organizational leverage.
Week 2 builds the AI-powered competitive intelligence system at production quality — with memory, self-evaluation, and a financial model that quantifies its impact for leadership. By the time you reach Day 30, the tools you mastered in Week 1 already have new capabilities you haven't explored. The Method ensures you never fall behind.
8Day
Build the CI System — Data Ingestion Layer
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Implement the first working layer of your competitive intelligence system: automated ingestion of public competitor signals into a structured Notion database.
  1. Open Make (or Zapier) and set up your first CI trigger: a scheduled scrape of your top competitor's G2 profile for new reviews posted in the last 7 days.
  2. Connect the trigger to your Notion CI database (schema from Day 4) — each new review becomes a row with signal type, source, date, and raw text.
  3. Use Claude Pro (or the OpenAI API via Make) to add an AI classification step: each incoming review is automatically tagged with your CI Signal Taxonomy categories from Day 4.
  4. Test the pipeline end-to-end: trigger it manually, confirm a review appears in Notion, confirm the AI classification is correct.
  5. Document any failures or misclassifications in a 'Pipeline Log' section of the Notion page.
Artifact
Day8-CI-Ingestion-Pipeline-v1.notion doc
At least one live competitor signal has been ingested into your Notion database, classified by the AI, and is readable by someone who wasn't in the room when you built it.
Reflection — 3 questions
  1. What broke first when you tried to implement the architecture you designed on Day 4 — and how did you fix it?
  2. How accurate was the AI's signal classification on the first pass? What patterns did you notice in its errors?
  3. What's the gap between the pipeline you have now and the one you'd be comfortable demoing to your VP of Sales?
Compliance note
This pipeline ingests public-source data only. Do not connect it to Gong call recordings or Amplitude product data — those require separate compliance review.
The difference between a CI system that exists in a Notion doc and one that actually runs is the difference between a plan and a product.
9Day
Add Self-Evaluation to the CI System
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Build a self-evaluation layer into the CI system so it flags low-confidence classifications and surfaces them for human review before they reach sales.
  1. Design a confidence-scoring prompt: for each incoming CI signal, the AI rates its own classification confidence on a 1-5 scale and explains why.
  2. Add a 'Review Queue' view to your Notion CI database — signals with confidence below 3 appear here for your manual review before being sent to sales.
  3. Test the confidence scoring on 10 real signals from your ingestion pipeline. Calibrate the threshold: what score level actually needs human review?
  4. Write a 'CI System Quality Protocol' — a one-page document describing how the self-evaluation layer works, what the Review Queue is for, and how often it should be checked.
  5. Update the pipeline documentation to include the self-evaluation step.
Artifact
Day9-CI-Self-Evaluation-Layer.notion doc
The Review Queue exists in Notion and contains at least 2 signals that the AI flagged as low-confidence. The Quality Protocol is specific enough that someone else could manage the queue without your help.
Reflection — 3 questions
  1. What types of signals did the AI consistently under-confident about — and what does that tell you about the limits of public-source CI?
  2. How does adding a self-evaluation layer change the trust calculus for sales reps using this system?
  3. What would a 'fully autonomous' version of this system look like — and what's the one thing you'd never want to automate away?
A system that knows what it doesn't know is more trustworthy than one that confidently outputs everything — and trustworthiness is what makes sales reps actually use it.
10Day
Build the Weekly CI Digest — Automated Draft
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Build the automated weekly CI Digest — a Make or Zapier workflow that pulls the week's classified signals from Notion and generates a formatted summary for sales and CS.
  1. Set up a weekly scheduled trigger in Make that queries your Notion CI database for all signals from the past 7 days.
  2. Pipe the week's signals into Claude (via API through Make) with a prompt that generates the CI Digest format you designed on Day 4.
  3. Configure the output to post to a Slack channel or send as an email — whichever your sales team actually reads.
  4. Run the workflow manually for this week's signals. Review the output against the digest template.
  5. Iterate on the prompt until the digest is something you'd be comfortable sending to the sales team without editing.
Artifact
Day10-CI-Weekly-Digest-Workflow.notion doc
A CI Digest for this week has been generated automatically and is formatted well enough to send to the sales team. The workflow runs without manual intervention beyond the initial trigger.
Reflection — 3 questions
  1. What did the AI include in the digest that you would have cut — and what did it miss that you would have added?
  2. How does the automated digest change your relationship to the competitive monitoring task? What do you do now that you didn't have to do before?
  3. What would make a sales rep forward this digest to their manager and say 'this is useful'?
An automated digest that sales actually reads is a revenue artifact — it shortens deal cycles and makes you the person who built the thing that closes deals.
11Day
Add the Job Posting Signal Layer
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Extend the CI system to ingest and interpret competitor job postings as strategic signals about their product direction and hiring priorities.
  1. Research the job posting feeds for your top 3 competitors (LinkedIn, Greenhouse, Lever, their careers pages) and identify which are scrapable via Make or Zapier.
  2. Add a job posting ingestion trigger to your CI pipeline — new postings from target competitors flow into the Notion database with a 'Hiring Signal' tag.
  3. Build a prompt that interprets job postings as strategic signals: what does hiring 5 ML engineers tell you about their roadmap? What does a new 'Head of Enterprise' role tell you about their go-to-market shift?
  4. Generate a 'Hiring Signal Analysis' for your top competitor's last 30 days of job postings.
  5. Add a 'Hiring Signals' section to the weekly CI Digest template.
Artifact
Day11-CI-Hiring-Signal-Layer.notion doc
You have a Hiring Signal Analysis for at least one competitor covering the last 30 days of postings, with at least 3 specific strategic inferences. The CI Digest template now includes a Hiring Signals section.
Reflection — 3 questions
  1. What was the most surprising strategic inference you drew from a competitor's job postings — something that changed how you think about their roadmap?
  2. How confident are you in the AI's strategic interpretations of hiring signals? Where did you override it?
  3. If you shared this hiring signal analysis with your VP of Product today, what would their reaction be?
Job postings are the most honest signal a competitor sends — they reveal what they're building before they announce it, and that's the kind of intelligence that wins deals.
12Day
Build the Python Ingestion Script
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Uncomfortable
Tools
Claude Sonnet 4.6Cursor (AI coding assistant)
Today's task
Goal: Write a Python script (using Cursor) that scrapes a competitor's public changelog or release notes page and outputs structured data for your CI database.
  1. Open Cursor and use its AI coding assistant to write a Python script that fetches a competitor's public changelog URL and extracts: release date, version number, feature names, and description text.
  2. Run the script in a Jupyter notebook environment. Debug with Cursor's AI until it produces clean output.
  3. Export the output as a CSV and import it into your Notion CI database manually (or via a Make trigger if you want to automate).
  4. Write a one-paragraph 'Script Documentation' block at the top of the notebook: what it does, what inputs it needs, what it outputs, and how to run it.
  5. Save the notebook as a versioned artifact.
Artifact
Day12-CI-Changelog-Scraper.ipynb other
The script runs without errors, produces structured output for at least one competitor's changelog, and the output is importable into your Notion CI database. The documentation block is complete.
Reflection — 3 questions
  1. How many times did you have to iterate with Cursor before the script ran correctly — and what type of errors came up most often?
  2. What's the difference between using Cursor to write this script and writing it yourself from scratch? What did you still have to understand to make it work?
  3. If you could automate one more data source with a similar script, which would have the highest CI value — and why?
Writing code with AI assistance is the gap-closing move that separates PMMs who design systems from PMMs who depend on engineers to build them.
13Day
Build the Pricing Intelligence Layer
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Add a pricing intelligence module to the CI system that tracks competitor pricing page changes and generates strategic interpretation for sales.
  1. Use a web monitoring service (Visualping free tier, or a Make webhook with a URL change trigger) to monitor competitor pricing pages for changes.
  2. When a change is detected, pipe the before/after diff into Claude Pro with a prompt that interprets the strategic implications: Did they raise prices? Add a tier? Remove a feature from a lower tier? Shift from seat-based to usage-based?
  3. Generate a 'Pricing Change Alert' template — a one-paragraph summary that sales can use immediately in competitive conversations.
  4. Test the module by manually simulating a pricing change for a competitor (use a cached version of their pricing page from a week ago).
  5. Add 'Pricing Intelligence' as a section in the weekly CI Digest.
Artifact
Day13-CI-Pricing-Intelligence-Module.notion doc
You have a working pricing change detection trigger and a Pricing Change Alert template that has been tested against a simulated change. The alert is specific enough that a sales rep could use it in a call without additional context.
Reflection — 3 questions
  1. What pricing change from a competitor in the last 6 months do you wish you'd caught faster — and how would this module have changed your response?
  2. How does the AI's interpretation of a pricing change compare to your own strategic read? Where did it miss the nuance?
  3. How does building a pricing intelligence module change how you think about your own company's upcoming pricing overhaul?
Pricing intelligence is the highest-stakes CI signal in B2B SaaS — a competitor's pricing change can shift a deal in 24 hours, and being the first to surface it is a revenue event.
14Day
Week 2 Milestone — The CI System is Alive
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Milestone
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Produce a CI System Demo Document — a structured walkthrough of the full system that you could present to your VP of Sales or VP Marketing today.
  1. Open Claude Pro and describe the full CI system you've built over Days 8-13: ingestion sources, classification layer, self-evaluation, digest format, job posting signals, changelog scraper, and pricing intelligence.
  2. Generate a 'CI System Demo Script' — a structured narrative that walks a non-technical stakeholder through the system in 10 minutes, covering: what it does, how it works, what signals it surfaces, and what it means for sales.
  3. Add a 'System Limitations' section — what the system cannot do, what it might get wrong, and how the human review layer (you) catches errors.
  4. Add a 'Next 30 Days' section — what you'll build next and what the system will look like at full maturity.
  5. Save the Demo Document as a shareable Notion page.
Artifact
Day14-CI-System-Demo-Document.notion doc
The Demo Document is a shareable Notion page that a VP could read in 10 minutes and understand what the system does, why it matters, and what it cost to build. The System Limitations section is honest and specific.
Reflection — 3 questions
  1. What's the one thing about the CI system that you're most proud of — and the one thing you're most nervous about showing to leadership?
  2. By Day 14, you've practiced The Method across 13 different tasks. What pattern have you noticed in your prompting that you didn't have on Day 1?
  3. What would it take to get the CI system to a point where it runs for a full week without you touching it? What's the one remaining dependency on your judgment?
A system that only you understand is a personal tool; a system you can demo to leadership is organizational infrastructure — and infrastructure gets resourced.
Milestone

Week 2 Complete: The CI System is Alive

  • Built a multi-source CI pipeline ingesting G2 reviews, job postings, changelogs, and pricing changes into a structured Notion database
  • Added a self-evaluation layer with a Review Queue — the system knows what it doesn't know
  • Automated the weekly CI Digest — a formatted summary that goes to sales without manual intervention
  • Wrote a Python changelog scraper using Cursor — closing the 'building as code' gap
Transitioning from Level 2 to Level 3 — first organizational-scale AI system is operational. Shift from personal productivity to team leverage is underway.
Week 3 builds the financial model that quantifies the CI system's impact, the leadership presentation that gets it resourced, and the Operators Playbook monetization architecture that turns your internal expertise into external revenue. The tools you mastered in Week 1 already have new capabilities you haven't explored — The Method ensures you never fall behind.
15Day
Build the CI System Financial Model
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
GPT-5ChatGPT Enterprise
Today's task
Goal: Build a three-scenario financial model that quantifies the revenue impact of the CI system for your leadership team.
  1. Open ChatGPT Enterprise and use it to generate the financial model framework — this keeps the financial modeling work within your company's enterprise AI environment.
  2. Define your three scenarios: Conservative (CI system reduces competitive deal loss rate by 5%), Base (10% reduction), Optimistic (20% reduction).
  3. For each scenario, calculate: deals affected per quarter, revenue impact at your average deal size, PMM time saved per week, and annualized ROI.
  4. Build the model in a Google Sheet or Notion table. Use real numbers where you have them (average deal size, win rate, number of competitive deals per quarter) and document your assumptions clearly.
  5. Add a 'Sensitivity Analysis' section: which assumption, if wrong, changes the model most?
Artifact
Day15-CI-System-Financial-Model.sheet sheet
The financial model has three scenarios, each with a calculated revenue impact and ROI. Every assumption is documented. The sensitivity analysis identifies the most critical assumption. A CFO could read it without a briefing.
Reflection — 3 questions
  1. Which assumption in your model are you least confident in — and how does that uncertainty change the story you'd tell leadership?
  2. What pattern have you noticed across the systems you've built so far — is there a type of problem you're consistently better at framing for AI than others?
  3. What's the minimum ROI scenario that would still get a 'yes' from your VP Marketing — and does your conservative scenario clear that bar?
A system without a financial model is a project; a system with a financial model is a business case — and business cases get resourced.
16Day
Build the Governance and Risk Artifact
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Produce a CI System Governance Document that addresses data sourcing, AI accuracy risks, and the human review protocol — making the system safe to present to legal and IT.
  1. Use Claude Pro to generate a Governance Document template for an AI-powered competitive intelligence system.
  2. Fill in the template with specifics: data sources (all public), AI models used (Claude via API, OpenAI via Make), data storage (Notion, no PII), human review layer (Review Queue), and update frequency.
  3. Add a 'Risk Register' section: list the top 5 risks (hallucinated intelligence, stale data, misclassification, data source unavailability, over-reliance by sales) with mitigation for each.
  4. Add a 'What We Do Not Do' section — explicitly stating that no customer data, call recordings, or internal product data flows through this system.
  5. Save as a shareable document you could send to your Head of Legal or IT Security without a cover note.
Artifact
Day16-CI-System-Governance-Document.notion doc
The Governance Document has a Risk Register with 5 risks and mitigations, a clear 'What We Do Not Do' section, and is specific enough that a non-technical legal or IT reviewer could assess it without asking you follow-up questions.
Reflection — 3 questions
  1. Which risk in your Risk Register is the one you're most worried about in practice — not in theory?
  2. How does writing a governance document change how you think about the systems you'll build in the future?
  3. What's one thing you'd want to add to this document before sharing it with your VP — and why haven't you added it yet?
Compliance note
This governance document is a starting point, not a legal opinion. Have it reviewed by your legal or compliance team before sharing externally.
A system that can survive a legal or IT review is a system that can be officially adopted — and official adoption is what turns a side project into organizational infrastructure.
17Day
Build the Leadership Presentation — Draft
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
GPT-5ChatGPT Enterprise
Today's task
Goal: Produce a first-draft leadership presentation that makes the case for the CI system — structured for a 15-minute slot with your VP Marketing or VP Sales.
  1. Use ChatGPT Enterprise to generate a presentation outline using the financial model (Day 15) and governance document (Day 16) as source material.
  2. Structure the presentation: Problem (what competitive intelligence looks like today), Solution (the CI system you built), Evidence (financial model scenarios), Risk Management (governance document summary), Ask (what you need from leadership to scale it).
  3. Write the speaker notes for each slide — what you'll say out loud, not what's on the slide.
  4. Identify the three hardest questions your VP will ask and write your answers.
  5. Save the presentation outline and speaker notes as a Notion document (slides can be built in Google Slides later).
Artifact
Day17-Leadership-Presentation-Draft.notion doc
You have a presentation outline with speaker notes for every slide and written answers to the three hardest anticipated questions. The 'Ask' slide is specific: a number, a resource, a timeline.
Reflection — 3 questions
  1. What's the one slide in this presentation that you're least confident about — and what would make you more confident?
  2. What's different about how you're thinking about 'presenting to leadership' now versus how you thought about it on Day 1?
  3. What would your VP Marketing say is the most important thing missing from this presentation right now?
The CI system is only as valuable as your ability to get leadership to resource it — the presentation is the system's go-to-market.
18Day
Stress-Test the Presentation with AI as Sparring Partner
Phase 2Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Uncomfortable
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Use Claude Pro as a hostile VP to stress-test your leadership presentation — find every weak argument before you're in the room.
  1. Paste your full presentation outline and speaker notes into Claude Pro.
  2. Instruct Claude to play the role of a skeptical VP Marketing who has seen many AI projects fail and is protective of the team's time and budget.
  3. Run the full presentation as if you're presenting live — Claude responds as the VP, asking hard questions and pushing back on weak claims.
  4. After the simulation, ask Claude to give you a structured debrief: which three arguments were weakest, which answer was most convincing, and what one change would most improve the presentation.
  5. Revise the presentation based on the debrief. Save the revised version.
Artifact
Day18-Leadership-Presentation-Revised.notion doc
You have a revised presentation that addresses at least 3 specific objections raised in the AI stress-test. The revision is substantive — not just wordsmithing.
Reflection — 3 questions
  1. What was the hardest question the AI-VP asked — and did your answer hold up?
  2. How does using AI as a sparring partner compare to practicing with a real colleague? What does each give you that the other doesn't?
  3. Are you ready to book the actual meeting with your VP? What's stopping you?
Every objection you handle in rehearsal is one you won't stumble over in the room — and stumbling in front of leadership is expensive.
19Day
Book the Meeting and Send the Pre-Read
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Uncomfortable
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Send the CI system pre-read to your VP and book the presentation meeting — this is the day the system goes from internal project to organizational conversation.
  1. Use Claude Pro to write a one-paragraph email introducing the CI system pre-read — what it is, why you built it, and what you're asking for in the meeting.
  2. Attach or link the CI System Demo Document (Day 14) as the pre-read. Do not send the full financial model yet — that's for the meeting.
  3. Send the email and book a 30-minute slot with your VP Marketing (or VP Sales, whoever is the right first audience).
  4. While you wait for the meeting, use the remaining session time to build a one-page 'CI System Summary' — a single Notion page that captures the system in under 5 minutes of reading.
  5. Save the email draft and the one-pager as artifacts.
Artifact
Day19-CI-System-One-Pager.notion doc
The email has been sent (or is scheduled to send). The meeting is booked or a calendar invite has been drafted. The one-pager exists and is shareable.
Reflection — 3 questions
  1. What did you feel when you hit send on the email — and what does that feeling tell you about your relationship to visibility?
  2. What's the best-case outcome of this meeting — and what's the realistic outcome you're actually planning for?
  3. What would you do differently about the CI system build if you knew from Day 1 that you'd be presenting it to leadership on Day 19?
Building a system is half the work; getting it adopted is the other half — and adoption starts with a calendar invite.
20Day
Operators Playbook — Monetization Architecture
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Design the full monetization architecture for Operators Playbook — paid tier structure, founding member offer, and revenue model.
  1. Use Claude Pro to analyze your newsletter's current state (400 subscribers, weekly cadence, AI for PMMs and product people) and generate three monetization architecture options.
  2. For each option, get: pricing structure, what's included at each tier, expected conversion rate from free to paid, and 12-month revenue projection.
  3. Pick the architecture that best fits your current audience size and content cadence. Document your reasoning.
  4. Design the 'Founding Member Offer' — a limited-time offer for your first 50 paid subscribers with a specific price, specific benefits, and a specific deadline.
  5. Save the monetization architecture and founding member offer as a Notion document.
Artifact
Day20-Operators-Playbook-Monetization-Architecture.notion doc
You have a chosen monetization architecture with a specific pricing structure, a Founding Member Offer with a price and deadline, and a 12-month revenue projection for the base scenario.
Reflection — 3 questions
  1. Which monetization option did you reject — and what does that rejection tell you about what you actually want Operators Playbook to be?
  2. What's the scariest number in the revenue projection — and is it scary because it's too high or too low?
  3. What will you do differently about building Operators Playbook now that you've designed the monetization architecture first?
A newsletter without a monetization architecture is a hobby; one with a founding member offer and a revenue model is a business — and you're building a business.
21Day
Week 3 Milestone — From System Builder to Strategist
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Milestone
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Produce a 'Strategic Position Document' that articulates your unique value proposition as an AI-native PMM — for internal use with leadership and external use in your market positioning.
  1. Use Claude Pro to synthesize everything you've built in the first 21 days into a Strategic Position Document.
  2. The document should articulate: what you've built (CI system, prompt library, launch workflow), what it's worth (financial model), what you know that others don't (your unique PMM + AI + devtools intersection), and where you're going (Operators Playbook, leadership positioning).
  3. Write two versions: a 200-word internal version for leadership conversations, and a 150-word external version for your LinkedIn 'About' section and Operators Playbook author bio.
  4. Identify the three most credible proof points from your first 21 days — the specific artifacts that demonstrate your AI-native PMM capability.
  5. Save both versions as a Notion document.
Artifact
Day21-Strategic-Position-Document.notion doc
You have two versions of your strategic positioning — internal and external — each under the word limit. The three proof points are specific artifacts with names and dates, not general claims.
Reflection — 3 questions
  1. What's the one thing you've built in the last 21 days that you're most confident talking about in a job interview or a podcast?
  2. What pattern have you noticed in the work you've done — is there a type of problem you're consistently drawn to, and is that your competitive advantage?
  3. Who in your network should read your external positioning statement today — and what would you want them to do after reading it?
You can't be seen as an AI-native PMM if you can't articulate what that means in your own words — this document is the foundation of everything that follows.
Milestone

Week 3 Complete: From System Builder to Strategist

  • Built the financial model and governance document for the CI system — ready for leadership presentation
  • Stress-tested the leadership presentation with AI as sparring partner and sent the pre-read
  • Designed the Operators Playbook monetization architecture with a Founding Member Offer
  • Articulated a Strategic Position Document in both internal and external versions
Operating at Level 3 — building organizational-scale systems, presenting to leadership, and beginning external market positioning.
Week 4 ships the Operators Playbook founding member launch, publishes your first AI-native PMM thought leadership piece, and assembles the master portfolio that makes your 30-day arc visible to the market. The skills you've built compound — every system you design from here gets faster and more powerful.
22Day
Write the Operators Playbook Launch Issue
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Write the launch issue of Operators Playbook that announces the paid tier — the issue that converts free subscribers into founding members.
  1. Use Claude Pro to generate a first draft of the launch issue using your monetization architecture (Day 20) as the brief.
  2. The issue structure: (1) a personal story about why you built the CI system and what it taught you, (2) the announcement of the paid tier and what founding members get, (3) the founding member offer with a specific deadline and CTA.
  3. Run the full 5-step loop: Describe the issue structure, Generate the draft, Validate against your editorial voice (does it sound like you?), Iterate on the personal story section (this is the hardest part for AI to get right), Review the full issue.
  4. Send the draft to one trusted subscriber for feedback before publishing.
  5. Save the final draft as a Notion document ready to publish.
Artifact
Day22-Operators-Playbook-Launch-Issue-Draft.notion doc
The launch issue is a complete, publish-ready draft. The personal story section sounds like you, not like an AI. The founding member offer has a specific price, specific benefits, and a specific deadline.
Reflection — 3 questions
  1. Where did the AI's draft sound most like you — and where did it sound least like you? What does that tell you about your editorial voice?
  2. What will you do differently about building Operators Playbook now that you've designed the monetization architecture and written the launch issue?
  3. What's the one thing you're most nervous about in the launch issue — and is that nervousness a signal to change it or a signal to ship it?
The launch issue is the moment Operators Playbook becomes a business — and the personal story is what makes people pay.
23Day
Build the Operators Playbook Course Outline
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Design the curriculum for the Operators Playbook course — the paid product that converts newsletter subscribers into students.
  1. Use Claude Pro to generate a course outline for a 4-week cohort course on AI-native product marketing for PMMs and product people.
  2. Structure the curriculum around the systems you've built in this Protocol: CI system design, prompt library productization, launch workflow automation, financial modeling for AI projects, and leadership presentation.
  3. For each module, define: learning objective, key concepts, hands-on exercise, and deliverable.
  4. Identify the three modules that are most unique to your experience — the ones no one else could teach with the same credibility.
  5. Write a 150-word course description for the Operators Playbook landing page.
Artifact
Day23-Operators-Playbook-Course-Outline.notion doc
The course outline has 4 modules, each with a learning objective, key concepts, hands-on exercise, and deliverable. The course description is under 150 words and is specific enough to put on a landing page today.
Reflection — 3 questions
  1. Which module in the course outline are you most excited to teach — and which one are you least confident you can deliver?
  2. What's the difference between the course you designed and the protocol you've been living for the last 23 days? What would you add or remove?
  3. Who is the ideal student for this course — and do you know 10 of them by name right now?
The course is the product that turns your expertise into recurring revenue — and the outline is the product spec.
24Day
Publish the First Thought Leadership Post
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Uncomfortable
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Publish a LinkedIn post or Operators Playbook issue that demonstrates your AI-native PMM expertise to the market — your first public proof-of-work artifact.
  1. Choose your topic: either 'How I built an AI-powered competitive intelligence system in 2 weeks' or 'The 5-step method I use for every AI task as a PMM.' Both are specific, credible, and valuable to your audience.
  2. Use Claude Pro to generate a first draft. Then run the full 5-step loop — pay special attention to the Validate step: does this sound like you? Does it teach something real?
  3. Add one specific, concrete example from your own work — a system you built, a failure you encountered, a result you measured. The AI cannot generate this; it must come from you.
  4. Publish to LinkedIn (and cross-post to Operators Playbook if appropriate).
  5. Save the published post URL and the draft as artifacts.
Artifact
Day24-Thought-Leadership-Post-Published.doc doc
The post is published (not just drafted). It contains at least one specific, concrete example from your own work that the AI could not have generated. The post URL is saved.
Reflection — 3 questions
  1. What was the hardest part of publishing — the writing, the editing, or hitting the button? What does that tell you?
  2. What's one thing you know about AI-native PMM work that you didn't see anyone else writing about in the last week?
  3. Who will you carry forward past this protocol — what will you publish next month, and the month after?
Publishing is the act that converts internal expertise into external positioning — and external positioning is what makes the Leap credible.
25Day
Build the Operators Playbook Landing Page Copy
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Write the full landing page copy for Operators Playbook's paid tier — the page that converts visitors into founding members.
  1. Use Claude Pro to generate landing page copy using your monetization architecture (Day 20), course outline (Day 23), and strategic positioning (Day 21) as source material.
  2. Structure the page: headline, subheadline, problem statement, solution, what's included, founding member offer, FAQ, and CTA.
  3. Run the 5-step loop with particular focus on the Iterate step: the headline and the founding member offer are the two highest-leverage elements — iterate on each at least 3 times.
  4. Test the copy on one person from your target audience (a PMM or product person you know). Ask them: 'Would you pay for this? What's unclear?'
  5. Save the final copy as a Notion document ready to publish to a landing page tool (Notion, Carrd, or similar).
Artifact
Day25-Operators-Playbook-Landing-Page-Copy.notion doc
The landing page copy is complete with all sections, the headline has been iterated at least 3 times, and at least one real person from your target audience has read it and given feedback.
Reflection — 3 questions
  1. What feedback from your test reader surprised you most — and did you incorporate it?
  2. What's the gap between the landing page you wrote and the one you'd see from a professional copywriter? Is that gap worth closing before launch?
  3. What's the one objection a potential founding member will have that your current FAQ doesn't address?
The landing page is the moment Operators Playbook becomes a product — and the headline is the moment a visitor decides whether to keep reading.
26Day
Publish the Operators Playbook Launch Issue
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Uncomfortable
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Publish the Operators Playbook launch issue to your 400 subscribers and open the founding member offer.
  1. Do a final review of the launch issue (Day 22) — read it out loud. Fix anything that doesn't sound like you.
  2. Set up the paid tier in your newsletter platform (Substack, Beehiiv, or equivalent) with the founding member pricing.
  3. Publish the launch issue and activate the founding member offer.
  4. Post a LinkedIn announcement linking to the launch issue — one paragraph, specific about what Operators Playbook is and who it's for.
  5. Track opens, clicks, and paid conversions for the first 24 hours. Save the metrics as a baseline.
Artifact
Day26-Launch-Issue-Published-Metrics.notion doc
The launch issue is published and live. The paid tier is active. The LinkedIn announcement is posted. 24-hour metrics are saved.
Reflection — 3 questions
  1. What happened in the first 24 hours — what surprised you about the response?
  2. What will you do differently for the next launch issue based on what you learned from this one?
  3. Who else in your organization should know that you launched a paid newsletter today — and what does it mean for how they see your role?
Shipping is the only proof that matters — everything before this was preparation.
27Day
Build the 30-Day Portfolio Document
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Assemble a master portfolio document that captures everything you've built in 30 days — the artifact that makes your AI-native PMM capability visible to any audience.
  1. Use Claude Pro to generate a portfolio document structure that presents your 30-day arc as a coherent narrative.
  2. For each major artifact (CI system, prompt library, launch workflow, financial model, governance document, leadership presentation, Operators Playbook launch), write a 2-3 sentence description: what it is, what problem it solves, and what it demonstrates about your capabilities.
  3. Add a 'Systems Built' section with a visual inventory of all systems and their current status (live / in progress / designed).
  4. Add a 'What I Learned' section — 5 specific lessons from the 30 days that you couldn't have written on Day 1.
  5. Save the portfolio as a shareable Notion page with a clean URL.
Artifact
Day27-30-Day-Portfolio-Document.notion doc
The portfolio document is a shareable Notion page with descriptions of all major artifacts, a systems inventory, and a 'What I Learned' section. You would be comfortable sharing the URL in a job application or a LinkedIn post.
Reflection — 3 questions
  1. Which artifact in the portfolio are you most proud of — and which one do you wish you'd spent more time on?
  2. What will you carry forward past the protocol — what's the one habit or system that you'll still be using in 6 months?
  3. Who else in your organization could benefit from a version of this protocol — and what would you change to make it relevant for them?
A portfolio is the artifact that makes invisible work visible — and visible work is what gets you promoted, hired, and followed.
28Day
Run the Exposure Diagnosis — Before and After
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Re-run the AI exposure diagnosis on your original 14 tasks and produce a before/after comparison that quantifies how your vulnerability has shifted in 30 days.
  1. Open Claude Pro and paste the original 14 tasks from your exposure diagnosis.
  2. For each task, re-score it on the 5 dimensions (Pattern vs Judgment, Processing vs Understanding, Analysis vs Gut Feel, Execution vs Decision, Output vs Ownership) — but this time, score your current capability, not the task's inherent automability.
  3. Build a before/after comparison table: original score, new score, delta, and one sentence explaining what changed.
  4. Identify the tasks where your score improved most — these are where the protocol had the highest impact.
  5. Write a 200-word 'Exposure Narrative' — the story of how your relationship to AI automation changed in 30 days.
Artifact
Day28-Exposure-Diagnosis-Before-After.notion doc
You have a before/after comparison table for all 14 tasks, a delta for each, and a 200-word Exposure Narrative. The narrative is specific — it names systems, artifacts, and moments, not general claims.
Reflection — 3 questions
  1. Which task's score surprised you most when you re-ran the diagnosis — and why?
  2. What's the one area where you're still more vulnerable than you'd like to be — and what would it take to close that gap?
  3. What will you do differently about AI and your career in the next 6 months based on what this before/after comparison shows?
Measuring change is the only way to know if the protocol worked — and the before/after comparison is the artifact that makes the case for your own growth.
29Day
Update LinkedIn and Send Two Real Outreach Messages
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Uncomfortable
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Overhaul your LinkedIn profile with your new positioning and send two specific outreach messages to people who should know about your work.
  1. Use your Strategic Position Document (Day 21) and portfolio (Day 27) to rewrite your LinkedIn headline, About section, and Featured section.
  2. Use Claude Pro to generate 3 headline options and 2 About section drafts — pick the ones that sound most like you and edit them.
  3. Identify two people to reach out to: one internal (a VP or senior leader who should know about the CI system) and one external (a PMM, founder, or investor in the devtools space who would find your work interesting).
  4. Write and send both messages — specific, short, and with a clear reason for reaching out (your Operators Playbook launch, the CI system, or a specific piece of their work you found relevant).
  5. Save the updated LinkedIn copy and both message drafts as artifacts.
Artifact
Day29-LinkedIn-Update-Outreach-Messages.notion doc
Your LinkedIn profile is updated with new headline and About section. Both outreach messages have been sent (not just drafted). The message drafts are saved.
Reflection — 3 questions
  1. What did you change about your LinkedIn headline — and what does the new version say about you that the old one didn't?
  2. What was harder: writing the internal outreach or the external outreach? What does that tell you about where your discomfort with visibility lives?
  3. Who else in your network should receive an outreach message in the next 30 days — and what would you say to them?
Positioning is only real when other people can see it — updating LinkedIn and sending real messages is the act that makes your 30-day arc visible to the market.
30Day
Day 30 Milestone — The Protocol is Complete
Phase 3Depth-investment: 90 minutes of thinking + refinement, even if the first draft completes in 8 minutes. Use the remaining budget for an experimentation pass, a second artifact variant, or a stress-test of the output.Milestone
Tools
Claude Sonnet 4.6claude.ai
Today's task
Goal: Write your 30-day retrospective and design your 90-day operating rhythm — the system that keeps you compounding after the protocol ends.
  1. Use Claude Pro to generate a structured 30-day retrospective using your portfolio document, before/after exposure diagnosis, and milestone summaries as source material.
  2. The retrospective should cover: what you built, what you learned, what surprised you, what you'd do differently, and what you're most proud of.
  3. Design your '90-Day Operating Rhythm' — a weekly schedule that maintains the habits you built: CI digest review, prompt library updates, Operators Playbook publishing, and one new system or experiment per month.
  4. Write a 'Letter to Day 1 Kenji' — one page, in your own words, telling yourself what the next 30 days will actually be like. This is for you, not for anyone else.
  5. Share the portfolio document URL publicly — in a LinkedIn post, in Operators Playbook, or both.
Artifact
Day30-30-Day-Retrospective-Operating-Rhythm.notion doc
The retrospective is written. The 90-day operating rhythm is a specific weekly schedule, not a list of intentions. The 'Letter to Day 1 Kenji' exists and is honest. The portfolio URL has been shared publicly.
Reflection — 3 questions
  1. What will you carry forward past the protocol — the one habit, system, or mindset that you'll still be using in a year?
  2. Who else in your organization could benefit from a version of what you built in the last 30 days — and what's the first step to making that happen?
  3. What's the first thing you'll ship publicly in the next 30 days — and what would it mean for your career if you actually ship it?
The protocol ends today, but the compounding doesn't — the 90-day operating rhythm is the system that makes everything you built keep growing.
Milestone

Day 30 Complete: AI-Native PMM, Fully Operational

  • Built a production-quality AI-powered competitive intelligence system with memory, self-evaluation, and an automated weekly digest
  • Launched Operators Playbook as a paid product with a founding member offer and a course outline
  • Delivered a leadership presentation on the CI system with a financial model and governance document
  • Published thought leadership, updated LinkedIn positioning, and sent real market outreach
Operating at Level 3 — designing and deploying organizational-scale AI systems, building external market positioning, and generating revenue from expertise.
The tools you mastered in Week 1 already have new capabilities you haven't explored. The Method ensures you never fall behind. Your 90-day operating rhythm is the system that keeps you compounding — one new system per month, one published piece per week, and the CI digest running without you. The market is moving fast. You're moving faster.
Milestones

Checkpoints along the way

DAY 7Week 1 Complete: From Stack to System
  • Audited and stress-tested all four existing AI systems — found and documented real failure modes
  • Added a memory layer to the battlecard Custom GPT using a structured Competitor Context Document
  • Designed a multi-source CI pipeline architecture ready for implementation in Week 2
  • Productized the 20-prompt library into a team-usable, documented system
Solidly at Level 2 — personal AI systems are hardened and documented. Beginning transition to Level 3 organizational leverage.
DAY 14Week 2 Complete: The CI System is Alive
  • Built a multi-source CI pipeline ingesting G2 reviews, job postings, changelogs, and pricing changes into a structured Notion database
  • Added a self-evaluation layer with a Review Queue — the system knows what it doesn't know
  • Automated the weekly CI Digest — a formatted summary that goes to sales without manual intervention
  • Wrote a Python changelog scraper using Cursor — closing the 'building as code' gap
Transitioning from Level 2 to Level 3 — first organizational-scale AI system is operational. Shift from personal productivity to team leverage is underway.
DAY 21Week 3 Complete: From System Builder to Strategist
  • Built the financial model and governance document for the CI system — ready for leadership presentation
  • Stress-tested the leadership presentation with AI as sparring partner and sent the pre-read
  • Designed the Operators Playbook monetization architecture with a Founding Member Offer
  • Articulated a Strategic Position Document in both internal and external versions
Operating at Level 3 — building organizational-scale systems, presenting to leadership, and beginning external market positioning.
DAY 30Day 30 Complete: AI-Native PMM, Fully Operational
  • Built a production-quality AI-powered competitive intelligence system with memory, self-evaluation, and an automated weekly digest
  • Launched Operators Playbook as a paid product with a founding member offer and a course outline
  • Delivered a leadership presentation on the CI system with a financial model and governance document
  • Published thought leadership, updated LinkedIn positioning, and sent real market outreach
Operating at Level 3 — designing and deploying organizational-scale AI systems, building external market positioning, and generating revenue from expertise.
Artifact index

What you will have built

DayArtifactPurpose
1The Method — Reference CardOne-page summary of the 5-step loop (Describe → Generate → Validate → Iterate → Review) to print and use daily.
1Day1-AI-Stack-Audit.notionStructured audit of all four existing AI systems with upgrade priorities.
2Day2-Battlecard-Failure-Mode-Registry.notionDocumented failure modes of the Custom GPT battlecard system with a Validation Protocol checklist.
3Day3-Competitor-Context-Template.notionMemory layer for the battlecard Custom GPT — filled for 3 competitors from public sources.
4Day4-CI-Pipeline-Architecture.notionFull architecture design for the multi-source competitive intelligence pipeline.
5Day5-PMM-Prompt-Library-v2.notionDocumented, rated, and tagged prompt library — team-usable with usage guides for top 5 prompts.
6Day6-Launch-Enablement-Deck-Workflow.notionReusable workflow and prompt template for generating a 7-slide sales enablement deck from a feature brief.
7Day7-Week1-Retrospective-Milestone.notionWeek 1 retrospective with artifact ratings, organizational impact ranking, and Week 2 intention.
8Day8-CI-Ingestion-Pipeline-v1.notionFirst working layer of the CI system — live competitor signals flowing into Notion.
9Day9-CI-Self-Evaluation-Layer.notionConfidence-scoring layer and Review Queue for the CI system.
10Day10-CI-Weekly-Digest-Workflow.notionAutomated weekly CI Digest workflow — formatted summary delivered to sales without manual intervention.
11Day11-CI-Hiring-Signal-Layer.notionHiring signal ingestion and strategic interpretation layer for the CI system.
12Day12-CI-Changelog-Scraper.ipynbPython notebook that scrapes competitor changelogs and outputs structured data for the CI database.
13Day13-CI-Pricing-Intelligence-Module.notionPricing change detection trigger and Pricing Change Alert template for the CI system.
14Day14-CI-System-Demo-Document.notionShareable walkthrough of the full CI system — ready to present to VP Sales or VP Marketing.
15Day15-CI-System-Financial-Model.sheetThree-scenario financial model quantifying the revenue impact of the CI system.
16Day16-CI-System-Governance-Document.notionGovernance document with Risk Register and 'What We Do Not Do' section — ready for legal and IT review.
17Day17-Leadership-Presentation-Draft.notionFirst-draft leadership presentation with speaker notes and anticipated Q&A.
18Day18-Leadership-Presentation-Revised.notionRevised presentation after AI stress-test — addresses 3 specific objections raised by the AI-VP simulation.
19Day19-CI-System-One-Pager.notionSingle-page CI system summary readable in under 5 minutes — sent as pre-read to VP.
20Day20-Operators-Playbook-Monetization-Architecture.notionPaid tier structure, Founding Member Offer, and 12-month revenue projection for Operators Playbook.
21Day21-Strategic-Position-Document.notionInternal and external versions of Kenji's AI-native PMM positioning statement.
22Day22-Operators-Playbook-Launch-Issue-Draft.notionPublish-ready launch issue announcing the Operators Playbook paid tier.
23Day23-Operators-Playbook-Course-Outline.notion4-module course curriculum and 150-word course description for the landing page.
24Day24-Thought-Leadership-Post-Published.docPublished LinkedIn post demonstrating AI-native PMM expertise — first public proof-of-work artifact.
25Day25-Operators-Playbook-Landing-Page-Copy.notionFull landing page copy for the Operators Playbook paid tier — tested with a real audience member.
26Day26-Launch-Issue-Published-Metrics.notionPublished launch issue, activated paid tier, and 24-hour metrics baseline.
27Day27-30-Day-Portfolio-Document.notionMaster portfolio document — shareable Notion page presenting the full 30-day arc.
28Day28-Exposure-Diagnosis-Before-After.notionBefore/after comparison of all 14 exposure diagnosis tasks with a 200-word Exposure Narrative.
29Day29-LinkedIn-Update-Outreach-Messages.notionUpdated LinkedIn profile copy and two sent outreach messages — internal and external.
30Day30-30-Day-Retrospective-Operating-Rhythm.notion30-day retrospective, 90-day operating rhythm, and Letter to Day 1 Kenji.
Resource links

Where to go next

Claude.ai
Your primary AI tool for Days 1-30. Use Claude Pro for all prompt engineering, document generation, and AI sparring sessions.
Open →
ChatGPT Enterprise
Use for financial modeling (Day 15) and leadership presentation drafting (Day 17) — keeps sensitive business modeling within your company's enterprise AI environment.
Open →
Make (Integromat)
Your automation platform for the CI pipeline — use for G2 ingestion triggers, pricing page monitoring, and weekly digest automation (Days 8-13).
Open →
Perplexity Pro
Use for real-time competitor research and public signal gathering — especially useful for job posting searches and release note discovery.
Open →
Cursor
Your AI coding assistant for the Python changelog scraper (Day 12) and any future code-level automation. Use it as a pair programmer, not a black box.
Open →
Notion AI
Your primary document workspace throughout the protocol. All artifacts are saved here. Use Notion AI for quick summarization and formatting tasks within documents.
Open →
Visualping
Free-tier web page monitoring for competitor pricing page change detection (Day 13). Set up alerts for your top 3 competitor pricing pages.
Open →
Beehiiv
Recommended newsletter platform for Operators Playbook paid tier setup (Day 26) — strong paid subscription features and analytics for newsletters at your stage.
Open →
Anthropic Prompt Engineering Guide
Reference for advanced prompt engineering techniques — useful when your prompts aren't producing consistent output and you need to debug them systematically.
Open →
OpenAI API Documentation
Reference for Make/Zapier integrations that call the OpenAI API — useful for the CI classification step (Day 8) and digest generation (Day 10).
Open →
Compliance notice

How this Protocol was produced

§

Transparency & Compliance

AI Disclosure
This Protocol was generated by an artificial intelligence system (Claude, Anthropic) based on your 19 assessment responses. The daily tasks, starter prompts, and recommendations are personalized but AI-generated. They are not professional career advice.
EU AI Act
This document complies with Article 50 of the EU AI Act (Regulation 2024/1689). Content generated by AI systems must be clearly labeled as such. This Protocol is AI-generated and personalized, not human-authored.
Data Basis
Recommendations are based on publicly available research, the AI Irreplaceability Protocol methodology, and your self-reported responses. The quality of recommendations depends on the accuracy of your inputs.
Methodology
The 5-Step Method (Describe -> Generate -> Validate -> Iterate -> Review) is a human-designed methodology integrated into the AI generation process. The daily task structure, difficulty progression, and milestone system were designed by human experts.