DE · EN
REPOSITION Workshop 2

Workflow Chain Templates

One brief, multiple roles, chained outputs. The framework for turning a single messy paragraph into specs, copy, and working code.

Get the templates

Workshop attendee? Enter the email you registered with for instant access. New here? You'll get the templates plus weekly AI tips.

Workshop attendees get instant access.

REPOSITION Labs opens after Workshop 6 with deeper content, recordings, and community.

The Workflow Chain

In any company, the same feature brief lands on different desks: the PM writes specs, the UX writer creates copy, the developer builds code. Each person needs different output from the same input.

The workflow chain pattern: pick a role, define the context, specify the task and format, then chain the output into the next role's prompt.

Feature Brief
PM Spec
UX Copy
Dev Code

Each step uses the same brief but a different role, task, and format. The output of one step can feed into the next.

The 4-Part Prompt Formula: Role, Context, Task, Format

Every prompt in a workflow chain has four parts:

Role — Who the AI is acting as ("You are a product manager writing specs for engineers")

Context — The information it needs (the feature brief, user feedback, requirements)

Task — What it should produce (user stories, acceptance criteria, copy, code)

Format — How the output should be structured (table, bullet points, specific sections)

The order can change. What matters is that all four are present.

Calibrate the Experience Level

A junior developer and a senior architect see the same problem differently. The experience level you set in the role determines the depth and sophistication of the output.

"5 years of experience" — mid-level patterns, practical solutions, moderate complexity.

"20 years of experience" — architectural thinking, edge cases, production-grade reasoning.

You don't need to fake your own experience. You're telling the AI what quality level to aim for.

The Feature Brief from the Workshop

We need a way for users to save articles they find interesting so they can read them later — like a reading list. It should work on mobile and desktop, maybe organize by topic. Users have been asking for this in feedback. They find stuff in their feed but lose it.

The Three Workflow Templates

Each template below takes the same brief but produces completely different output based on the role.

Template 01 Product Manager

The PM Spec Generator

Live demo: Turning a messy paragraph into structured product specs
You are a product manager writing a spec for engineers. Here is a feature brief from a stakeholder: [paste your feature brief here] Produce: 1. A clear problem statement 2. 3-5 user stories in standard format 3. Acceptance criteria for the core story 4. Edge cases to consider Use a standard product spec format. Be specific.
This turned a single vague paragraph into a structured spec with user stories, acceptance criteria, and edge cases in about 15 seconds. In a real company, this takes 2-3 hours.
Template 02 UX Writer

The UX Copy Machine

Live demo: Generating feature copy, changelog, and error messages
You are a UI/UX writer at a tech company. Here is a feature brief: [paste your feature brief here] For this feature, produce: 1. Feature name options (3 choices) 2. A one-line changelog description 3. An in-app announcement (under 100 words) 4. Error messages for common failure cases 5. Tone guidelines for this feature's copy Give 3 options for the feature name and announcement.
Notice this uses the same brief as the PM template but produces completely different output. The UX writer cares about naming, tone, and user-facing text, not stories or acceptance criteria.
Template 03 Developer

The Developer Workflow

Live demo: Claude Code building a FastAPI backend from the feature brief
You are a senior developer with 20 years of experience in [Python / your stack]. Here is a feature brief: [paste your feature brief here] Generate a [FastAPI / your framework] backend with: - [save article endpoint] - [list saved articles endpoint] - [remove article endpoint] - Data model with validation - Routes, schemas, and basic tests [Keep it in one file for prototyping / structure it as a real project]
This template works best in Claude Code (the terminal tool), where it can create actual files. In the workshop, it generated a working FastAPI backend with models, routes, schemas, and tests from the single brief.

The Chaining Technique

How to connect outputs from one role into the next.

Template 04 Workflow Chain

The Multi-Role Chain

From the workshop: how one person can act as PM, UX, and Dev
You are a [role: startup founder / solo dev / team lead]. I have a product idea: [describe your product or feature in 2-3 sentences] Run this through three roles sequentially: Step 1 - As a [Product Manager]: Create a product brief with user stories and acceptance criteria. Step 2 - As a [UX Writer]: Take the PM output and create feature naming, in-app copy, and error messages. Step 3 - As a [Senior Developer]: Take both outputs and create a technical implementation plan. After all three, summarize the key decisions and any conflicts between the roles.
This is the solo-developer startup pattern: one person, three hats. The AI acts as each role and the outputs feed into each other. For complex products, break this into three separate prompts instead of one.
Template 05 Strategy

Break a Multi-Role Prompt into Steps

From the live review: when a single prompt tries to do too much
I'm building a [SaaS product for X audience]. I need to go from idea to implementation. Instead of one big prompt, guide me through these steps one at a time: Step 1: [Founder] — Build a mission and vision for this product. What problem does it solve? Who is it for? (After I review Step 1, continue to Step 2) Step 2: [Product Manager] — From that mission, create a product brief. What are the core features? Who are the competitors? (After I review Step 2, continue to Step 3) Step 3: [CTO / Architect] — From the product brief, create the system architecture. What's the tech stack? What are the key components? Start with Step 1. Do not proceed until I confirm.
The key phrase is "Do not proceed until I confirm." This prevents the AI from running ahead. You review each step, make adjustments, then move on. Quality over speed.
Template 06 Everyone

The Sparring Partner Prompt

From the workshop: using AI to accelerate learning, not just copy-paste
You are a [role] with [years] years of experience in [domain]. I'm working on [your situation]. My experience level is [junior / mid / senior]. Give me [the output you need]. After generating the output, explain: 1. Why you structured it this way 2. What a [more junior / more senior] person might do differently 3. One thing I should learn to do this without AI help
The "explain why" part turns AI from a copy-paste machine into a learning tool. When the AI explains its reasoning, you build understanding that compounds over time.
This site uses essential cookies only. Privacy Policy