Written by Mattia Gottardello – Full-stack Developer in AzzurroDigitale
In the previous episode — “Beyond the Hype” — in the previous episode, “Beyond the Hype” — we dismantled the mystique surrounding LLMs: they’re not oracles, but statistical engines that predict the next word. Good. Now that we know what an LLM is, we need to learn how to drive it. Because the difference between a mediocre answer and an output that saves you two hours of work isn’t the model itself — it’s how you ask.
Same AI, opposite results
Prompt Engineering is not black magic — it’s process optimization. The same technology can produce radically different results depending on how the request is structured.
A poorly written prompt is like assigning a task to an intern on their first day with zero context: you end up with generic, off-tone text that has to be rewritten from scratch.
A well-structured prompt is like working with a senior colleague: they receive the instructions, understand the context, and deliver something usable on the first try.
Translated into concrete business terms:
- Sales emails that match your tone of voice instead of sounding robotic.
- Reports and document summaries already formatted according to your internal standards, without manual rework.
- Consistent outputs at scale, independent of daily mood swings or rushed execution.
The ROI is easy to calculate: less time spent correcting, more time spent making decisions.
Easy to get wrong
Let’s take a common daily task: writing an email to a customer about a delivery delay.
Bad Prompt (lazy zero-shot)
Result: an anonymous, overly formal text written in stereotypical corporate English, signed “The Team.” Completely unusable.
Proper Prompt (structured)
Same model, completely different outcome.
The rule is simple: give the AI the same context you would give a human being. Role, objective, constraints, output format. If omitting a detail would confuse a colleague, it will confuse the model too.
Deep Tech: Technical Insights and Best Practices
When moving from chat interfaces to API integrations, the game changes. The first key distinction is architectural:
- System Prompt — The persistent “rules of engagement” for the agent: who it is, what it can and cannot do, and the required response format. Defined once.
- User Prompt — The specific request from the end user, changing with every call.
Two techniques I use daily in producion:
Few-Shot Prompting — Include 2–3 input/output examples directly in the payload. The model learns the pattern far better than through abstract descriptions alone. This is essential for custom taxonomies, structured data extraction, domain-specific classifications.
Chain-of-Thought — you force the model to reason before answering. Typical instruction: “Analyse the steps in a reasoning field, then return the final result in output”. Drastically reduces hallucinations on logical tasks, at the cost of a few extra tokens. When the output has to be JSON consumed by other code, this pattern makes the difference between a reliable integration and a production bug at 3 a.m.
Conclusion and Next Episode
Today, the best code is no longer made only of Python, TypeScript, or Go: it also includes well-engineered natural language instructions. Treat prompts the same way you treat code: version them, test them, document them.
Next episode —“LLMs in Production: Costs, Security and Integration“: how to bring these models onto company servers without blowing the budget or risking data leaks. We’ll talk about token economics, RAG, self-hosting and privacy. Stay tuned.
Frequently Asked Questions (FAQ)
What is prompt engineering?
It is the practice of structuring requests to an artificial intelligence model to get useful, consistent outputs. It is not magic but process optimisation: the quality of the answer depends on how the request is set up.
How do you write a good prompt?
By giving the AI the same context you would give a colleague: role, goal, constraints and output format. If an omitted detail would confuse a person, it will confuse the model too.
What is the difference between a system prompt and a user prompt?
The system prompt defines the assistant’s persistent rules: who it is, what it can do and in what format it must answer. The user prompt is the user’s individual request, which changes with every call.
What is few-shot prompting?
It is a technique in which you include 2-3 input and output examples in the prompt: the model learns the pattern better than from an abstract description. It is useful for classifications, taxonomies and structured data extraction.
What is chain-of-thought for?
It asks the model to spell out its reasoning before the final answer. It reduces hallucinations in logical tasks, at the cost of a few extra tokens, and is useful when the output has to be used by other code.