GESHDO

The Playbook

Methodology & Prompt Templates

How we build with AI — from our core engineering methodology to battle-tested prompts you can copy directly into your workflow.

Core Methodology

Spec-Driven Development

Start by writing down your intent. Describe expected behavior, constraints, and success criteria before writing a single line of code. The specification isn't just documentation—it's the contract between product, design, and engineering.

  • Code fulfils the spec, not individual interpretations
  • Spec guides implementation & PR reviews
  • Changes start with spec updates
  • Prevent drift between intent and code

Comparison

Why Methodologies Matter

Understanding the difference between structured engineering and free-form generation is key to success.

Methodology Philosophy Focus Outcome
Spec-Driven Dev "Think before you code" Intent, Constraints, Contracts Predictable, Maintainable, Scalable
Regular Development "Code first, refine later" Implementation details, Logic Functional, but prone to drift
Vibe Coding "Flow state, intuition first" Speed, Experimentation, "Feel" Creative, Fast, but chaotic

Guidelines

The Rules of Engagement

What to do

Use AI as a Coding Buddy

Treat it like a junior dev. It helps, but you result is your responsibility.

Think AI First

Identify where AI can save you time before you start typing.

Experiment & Learn

New models drop weekly. Try them out and find what works best for your workflow.

Plan Your Work

Think ahead. Don't burn credits on rewrites because you didn't have a plan.

What NOT to do

Don't Blindly Copy-Paste

Understand the code. If you can't explain it, don't ship it.

No Secrets in Public AI

Never feed confidential data (keys, PII, internal configs) to public models.

Don't Assume It's Right

AI hallucinates. Double-check logic, security, and edge cases.

Don't YOLO

Jumping in without a plan wastes time and API credits.