Market Discovery & Empathy
Summarize these 500 customer reviews.
An executive summary hides the bleeding. If users hate the checkout button, the AI will say "Some users experience payment friction." You cannot build a backlog from a summary.
Act as a forensic data analyst. Ingest this raw feedback. Do not summarize. Cluster the data by "Severity" and "Churn Risk". Extract the exact top 3 bugs causing the highest drop-off rate, quoting the users directly.
Act like a typical millennial banking customer and tell me what features you want.
The AI hallucinates a stereotype based on Reddit data. It will ask for "crypto and fast UI." It is not representing YOUR actual angry users.
Using ONLY the highest-severity cluster from the previous prompt, construct a "Synthetic Persona" representing our most frustrated user. Give them a name, context, and a primary grievance. Do not invent details outside the data provided.
Would you like a dark mode feature?
The AI will politely say yes. You are leading the witness. You must force the AI to be hostile and protective of its time to simulate a real B2B interview.
Stay in character as the frustrated Persona. I am pitching a 'Dark Mode UI' Epic. Respond with extreme skepticism based on your recent transaction trauma. Score my pitch out of 10 for actual value, and tell me what I SHOULD be building instead.
Compare our app to our biggest competitor.
A generic Wikipedia-level comparison. It tells you what everyone already knows. We need actionable product gaps.
Act as an aggressive disruptor Product Manager. Build a feature-parity matrix comparing us to [Competitor]. Do not list similarities. Identify the "Blue Ocean" gapβthe specific feature they are missing that costs them high-net-worth users.
Why do people uninstall apps?
You asked a generic question, you got a generic blog post. We need to correlate our specific data to churn triggers.
Analyze our support ticket data again. Identify the "Silent Churn" indicatorsβissues that users mention in passing before abandoning the app entirely, rather than complaining loudly about. Output a risk-mitigation Epic for the backlog.
Value, Revenue & Prioritization
Here are 5 Epics. Prioritize them for me.
The AI guessed based on training bias. It didn't ask for business value or effort. Never let an AI make unconstrained roadmap decisions.
Act as a Lean Portfolio Manager. I will provide 5 Epics. Force me to score (1-10) for: User/Business Value, Time Criticality, Risk Reduction, and Job Size. Then, calculate the WSJF score for each and output a ranked Markdown table.
What happens if we delay Epic 1?
You get a generic "users might be unhappy." You need hard financial projections to defend your backlog to the CFO.
Using the WSJF table, calculate a hypothetical "Cost of Delay" (CoD) in dollars for delaying the #1 ranked Epic by 4 weeks. Assume our app generates $100k MRR. Draft a 3-bullet executive summary defending why Epic 1 cannot be delayed.
Write a polite email to the CEO telling him we aren't building his pet project.
A polite email gets you fired. A data-backed counter-proposal gets you promoted.
Draft a data-backed counter-proposal for the CEO. Use the WSJF matrix and the Cost of Delay calculation to prove that prioritizing his pet project (Epic 4) will cost the company MRR and increase churn among our highest-severity users.
Are there any risks with Epic 1?
The AI will say "There might be software bugs." You need specific industry compliance risk mapping.
Act as a strict Chief Compliance Officer for [Your Industry]. Review our top-ranked Epic. Identify the hidden regulatory landmines (e.g., GDPR, HIPAA, SOC2 compliance risks) that the dev team will likely overlook. Generate 2 mandatory compliance user stories.
Does this backlog look good?
Sycophancy again. The AI says yes. You need to ensure your backlog actually hits the board's targets.
Our Q3 Corporate OKR is "Increase User Retention by 15%". Audit our prioritized Epic backlog against this specific OKR. Flag any Epics that are "Vanity Projects" failing to directly drive this metric. Suggest replacements.
Orchestration & Execution
Break this Epic into user stories for the dev team.
The AI will slice horizontally: "Build DB," "Build API," "Build UI." This creates technical silos and zero usable value until the end of the sprint.
Slice this Epic VERTICALLY into exactly 6 User Stories. Every story MUST deliver end-to-end user value crossing all architectural layers. Strictly adhere to the INVEST principle. Flag any story that seems too large for a standard 2-week sprint.
Write acceptance criteria for Story 1.
It provides a generic bulleted checklist. Developers will misinterpret it, leading to bugs.
Generate exhaustive Behavior-Driven Development (BDD) criteria for Story 1. Use strict Gherkin format (Given / When / Then). Ensure there are at least 2 Happy Paths and 2 Negative Paths mapped out.
Did you miss anything in these stories?
The AI will reassure you that it did a great job. It won't find the edge cases until you force it to think like a hacker.
Act as a chaotic user trying to break the system. Review the acceptance criteria you just wrote. Identify 3 bizarre edge cases (e.g., rapid double-clicking, network drop during save, invalid character inputs) and write new criteria to prevent them.
Are there any dependencies here?
It will give obvious answers like "You need a database." You need to anticipate blockers before Sprint Planning.
Act as a Lead Systems Architect. Review the 6 generated stories. Map out the hidden technical dependencies between them. Identify which story poses the highest architectural risk and must be spiked/prototyped first.
How many story points is this?
AI cannot accurately guess story points for your specific team. It hallucinates numbers.
I do not want story points. I want a Complexity Audit. Compare these 6 stories against each other based strictly on logic complexity and integration risk. Output a relative sizing chart (S, M, L) to help my dev team baseline their estimates during planning.
Quality, Telemetry & Partnership
Review your work to make sure it's perfect.
The AI will literally just reply "It is perfect." You must trigger persona switching.
Forget your previous persona. Act as the world's most pedantic, aggressive Lead QA Engineer. Interrogate the acceptance criteria. Find 3 massive loopholes a junior developer will exploit to build the wrong thing. Rewrite to close the loopholes.
Analyze this sprint data. Why did velocity drop?
The AI states the obvious: "Velocity dropped because fewer points were finished." You need systemic root-cause analysis.
Act as an Elite Agile Coach. Correlate the Cycle Time data against Code Review Duration and Bug Rates. Diagnose the hidden systemic bottleneck (e.g., Is QA starving because devs batch commits?). Output a root-cause hypothesis.
Give me 3 questions to ask in the retro.
You get generic "What went well / What didn't" questions. This leads to boring, useless retrospectives.
Based on the bottleneck diagnosed in the previous step, design a targeted 30-minute Retro agenda. Generate 3 provocative, psychologically safe questions designed to help the engineering team realize the batching issue themselves without feeling attacked.
Write release notes for the stories we finished.
The AI writes technical jargon that customers don't care about ("We updated the API endpoints").
Translate these completed Jira stories into Customer-Facing Release Notes. Do not mention APIs or databases. Focus 100% on the unlocked business value and how this makes the user's life easier today. Match the tone of an Apple product launch.
Write a sprint summary for the board.
The board does not care about your sprint velocity or story points. They care about money and risk.
Draft the Executive ROI Report for the Sprint Review. Map the delivered features directly back to the Q3 OKRs. Calculate the estimated Cost of Delay we successfully mitigated by delivering this on time. Keep it under 200 words.
Data Repository (BYOC)
Select your industry below. Use this data for Session 1 and Session 2.