RUNLOYAL ENGINEERING

AI Workflow
Team Onboarding Guide

How we use AI to ship features faster — without cutting corners. Read this once before you start your first feature.

What is this workflow?
A 7-stage process where every stage has a ready-made AI prompt. You fill in the context, the AI does the heavy drafting, you review and approve, then move to the next stage.
📋

It's a process

7 stages from raw feature idea to live in production. Each stage has a clear input, a clear output, and a done checklist.

🤖

AI does the drafting

You paste the prompt into Claude, GPT-4o, or any LLM. It generates a draft. You review it. You own the result.

📁

Everything is written down

Each stage produces a .md file. No lost Slack decisions. The whole feature is traceable.

How to use it
For every new feature, follow these steps in order. Don't skip stages.
1

Open the stage file for where you are

Each stage has its own .md file: stage-01-feature-clarification.md, stage-02-check-existing-flow.md, and so on up to stage 07. Find the one matching your current stage.

2

Copy the prompt block

Inside the file, find the section labelled 🤖 AI Prompt. Copy everything inside the code block — that's the prompt you'll paste into the AI.

Look for the ``` block under "🤖 AI Prompt" — copy everything between the backticks
3

Fill in the [PLACEHOLDERS]

Every prompt has clearly marked spots like [PASTE STAGE 01 RESULT HERE] or [FEATURE REQUEST]. Replace each one with the real information before sending to the AI.

For stages 03–07: always paste the output from the previous stage into the next prompt
4

Paste into your AI model and run it

Open Claude (claude.ai) or ChatGPT or any LLM. Start a new chat. Paste the filled-in prompt. Send it. Wait for the full response.

Recommended: Claude for design stages (03, 04) · Claude or GPT-4o for code (05)
5

Review the output — you own it

Read every line. The AI drafts, you decide. Fix anything that doesn't match reality. Add anything it missed. This is not optional — AI output must be reviewed by a human who knows the context.

6

Save the output as a .md file

Create a folder for the feature: features/2025-01-15-pet-reminders/. Save the output as the appropriate file, e.g. 01-feature-definition.md.

This file becomes the input for the next stage's prompt
7

Tick the done checklist, then move on

Each stage file ends with a ✅ Done When checklist. Every box must be checked before proceeding to the next stage. If a box can't be checked, resolve the blocker first.

Who owns each stage
Every stage has a primary owner. The AI assists, but the owner is accountable for the output quality.
Stage Name Primary Owner AI Involvement Key Output
01 Feature Clarification Product Drafting aid Feature definition + user stories
02 Check Existing Flow Tech Lead Analysis aid Flow diagram + dependency map
03 Impact Analysis Tech Lead AI Risk identification Impact list + risk & mitigation plan
04 Feature Design Senior Eng AI Architecture drafting Backend + DB + frontend design doc
05 Development Engineer AI Code generation + tests Working code + PR summary
06 Testing & Regression QA / Eng AI Test case generation Test report + security audit
07 Release & Deployment Tech Lead AI Release notes + runbook Release notes + rollback plan
Golden rules
What to always do and never do when using this workflow.

Always paste the previous output

Every prompt asks for prior stage results. Include them. The AI has no memory — without context it guesses.

Don't skip stages

Going from idea to code without Impact Analysis leads to rework. Every stage catches something the next one shouldn't have to fix.

Review every AI output

AI drafts, you decide. Read it line by line. Fix inaccuracies. Add what it missed. You own the output.

Don't treat AI output as final

AI doesn't know your codebase, your customers, or your constraints. It gives a head start — not a finished answer.

Save every output as a .md file

These files are the project memory. Name them consistently and keep them in the feature folder.

Don't move forward with open blockers

The done checklist is a gate, not a suggestion. Unresolved risks in stage 03 will become bugs in stage 05.

Common questions
Things the team asked when we introduced this workflow.

Which AI model should I use?

Any instruction-following model works — Claude, ChatGPT, Gemini. For design stages (03, 04) Claude tends to produce more structured output. For code generation (05) Claude or GPT-4o both work well. Use what you have access to. The prompts are model-agnostic.

What if the AI output is wrong or missing things?

Fix it. The AI is a fast first draft, not a source of truth. If you notice a gap, add it to the file before moving on. The done checklist will catch most missing items.

Do I have to do all 7 stages for a tiny bug fix?

No. Use judgment. A one-line bug fix doesn't need a risk register. A new user-facing feature or anything touching shared infrastructure should go through all 7. If in doubt, start at stage 03 (Impact Analysis) as a minimum.

Where do I save the .md files?

Create a folder in the repo under features/ named with the date and feature slug, e.g. features/2025-01-15-pet-reminders/. Save each stage output there with the matching filename from the master guide.

What if a stage is blocked waiting for someone else?

Document the blocker in the output file under "Open Questions" and tag the owner in your project tracker. Do not proceed to the next stage until it's resolved — the whole point of early stages is to catch blockers before they become expensive.

Stage quick reference
Bookmark this. Each row = one stage file on your system.

📂 stage-0X-[name].md

01 Feature Clarification → definition + user stories
02 Check Existing Flow → flow diagram + dependencies
03 Impact Analysis → impact list + risk plan
04 Feature Design → backend + DB + frontend design
05 Development → code + tests + PR summary
06 Testing & Regression → test report + security audit
07 Release & Deployment → release notes + deploy runbook + rollback plan