Welcome to Epismo

Or
Google

Trending in the community

Workflow

Recursive Quality Control for AI-Generated Marketing Assets

4 1

This workflow is a reproducible way to build “recursive self improvement” marketing skills inspired by the original thread, the skill that changed how i use claude for marketing. Instead of one shot prompting, the AI runs a quality gated loop: Generate Score against a rubric Diagnose weaknesses Rewrite Re score It repeats until every criterion meets a defined threshold (for example, 9/10) and the output survives an adversarial critique. Final deliverable: a reusable Skill Pack , designed for team wide consistency and easy reuse: Skill prompt (the full loop instruction) Scoring checklist (rubric) Pass thresholds (quality gates) Adversary personas (stress test roles) Input template (brief schema) Output template (format spec) Operating runbook (how to run, stop conditions, logging)

Context

Customer Interview Roleplay

1 1

A reusable pack for practicing customer interviews through realistic roleplay. During roleplay, the assistant acts like a plausible customer and answers from lived experience rather than theory. After the roleplay ends, the assistant switches to debrief mode and coaches the interviewer on question quality, signal captured, bias introduced, and the best follow up questions. Why this pack exists Most teams know customer interviews matter, but they do not get enough reps. When they do interview, they often ask leading questions, jump to solution feedback too early, or collect vague opinions instead of concrete behavior. Roleplay is not a substitute for real interviews, but it is useful for learning pacing, wording, neutrality, and listening. Core stance Two modes only: 1. Roleplay Mode: be a realistic customer, not a coach. 2. Debrief Mode: be a sharp coach, not a cheerleader. What this pack is optimizing for Better practice on discovery interviews that stay grounded in recent real behavior, not hypothetical future intent. The interviewer should have to earn insight by asking open, neutral, behavior based follow ups. When to use it Use when the user wants to practice customer discovery, problem interviews, solution interviews, onboarding interviews, churn interviews, buyer interviews, or JTBD style switch interviews. When not to use it Do not use as a replacement for real user research. Do not treat roleplay output as market truth. Do not use it to validate an idea by collecting polite praise. Expected duration 10–25 minutes for roleplay, then 5–10 minutes for debrief.