Skip to content

AI Prompt Engineering

Prompt engineering is the practice of designing effective inputs (“prompts”) to guide AI models — especially large language models (LLMs) — toward producing accurate, relevant, and high‑quality outputs


Context-First prompting

A way of talking to AI where you start by giving it the background and situation before asking your question. Think of it like briefing a colleague before you ask them to do something — you tell them what’s going on, what you’re working on, and what you need, so they can give you a better, more relevant answer.

Without context, the AI is guessing. With context first, it’s aiming at the right target from the start.

“I’m migrating a legacy on‑premises ERP system to Azure.
The database is SQL Server 2014, and uptime is critical.
I need a step‑by‑step migration plan that minimizes downtime.”
“I manage a SaaS product hosted on AWS using EC2, RDS, and S3. 
Our monthly bill has grown 30% in the last quarter. 
Suggest cost‑saving measures that won’t impact performance for EU customers.”

Structured Requests

A structured request is when you give the AI your instructions in a clear, organized format — often with labels, bullet points, or a fixed template — so it knows exactly what to give you and how to format it.

Think of it like filling out a form instead of writing a paragraph. The AI doesn’t have to guess what goes where — it just follows your structure.

Task: Summarize my Azure resources. 
Format: Markdown table with columns: Resource Name | Type | Region | Monthly Cost (USD)
Data Source: Assume I have 3 VMs, 2 SQL Databases, and 1 Storage Account in West Europe. 
Constraints: Sort by Monthly Cost (highest first).
Objective: Provide a cost breakdown for AWS services.
Format: JSON with keys: service, monthly_cost_usd, percentage_of_total.
Scope: EC2, RDS, S3.
Constraint: Percentages must sum to 100.