This article was originally published on LinkedIn.

When NetSuite first introduced SuiteQL, it didn't make headlines. Most users stuck with saved searches or reports and never thought twice about writing queries.

For those of us who learned it — or who already knew SQL — SuiteQL became a sort of superpower. It let us reach deeper into the system, access data others couldn't, and solve problems the UI couldn't touch.

I think prompt engineering is the same kind of skill — but this time, the system isn't relational. It's generative.

Prompting Feels a Lot Like Programming

As I've continued to experiment and work with LLMs, prompting has started to feel like a strange new form of programming.

You're writing instructions — but not in a formal language. You're defining structure and flow — but in plain English. And sometimes, it feels like debugging with a ghost: you fix one part of a prompt and something unrelated breaks somewhere else.

There have been many times when I've had to stop and remind myself: I'm not writing code.

And yet, the parallels are there — problem decomposition, flow control, variable substitution, input validation, and output formatting.

The big difference is that prompting requires intent-awareness. You're not just telling a machine what to do; you're describing what success should look like — and what failure must not look like. And with prompt engineering, you often leave the how up to the model.

That's a fundamentally different mindset from coding — and it takes time to get used to.

The Frustrating Part: It's Non-Deterministic

Unlike code, you can run the exact same prompt multiple times and get different results.

That can be unsettling for people coming from traditional development backgrounds. We're used to precision, repeatability, and control — not probability, tone, and nuance.

But that's also part of what makes LLMs powerful. You're not just executing logic — you're exploring a range of interpretations. And with the right structure, guardrails, and context, that variability can actually lead to better results.

Still, it requires a shift in how we think. Prompting isn't about scripting — it's about shaping.

Why It's the New SuiteQL

SuiteQL didn't make saved searches obsolete. It just gave developers and analysts a deeper way to work with the same data — a lower-level interface for people who wanted precision and control.

I think prompt engineering works much the same way. Tools like NetSuite's AI Connector, built-in assistants, and SuiteScript LLM APIs will eventually abstract a lot of prompting away.

But knowing how to speak the model's language directly will still matter — and it will set apart the developers, analysts, and consultants who can make AI do exactly what they need it to do.

It's the difference between using AI and understanding AI.

A Skill for Everyone

One of the biggest differences between prompt engineering and traditional programming — or even SuiteQL — is who it's for.

You don't need a head for code to get good at prompting. You need a head for business.

Prompt engineering rewards people who understand context, process, and intent — people who know what they're trying to achieve, why it matters, and what success should look like.

In that way, it's far more accessible than something like SQL. You don't have to memorize syntax or debug line numbers. You just have to think clearly, ask good questions, and iterate.

I've already seen people who've never written a line of code get incredible results because they know how to think through a business problem — not because they're "technical," but because they're curious, analytical, and outcome-oriented.

That's why I think prompt engineering is such a transformative skill. It's not just for developers — it's for everyone who wants to turn ideas into action.

Beyond NetSuite

This skill isn't limited to one platform. Whether you're working with NetSuite, SAP, Workday, or QuickBooks — or with middleware like Boomi or Celigo — the underlying concept is the same: you're bridging structured systems (ERPs, CRMs, databases) with probabilistic reasoning engines (LLMs).

The ability to do that well — to prompt precisely, safely, and contextually — will become an enterprise skill in its own right.

In that sense, prompt engineering for ERP systems may become what SQL once was for databases: a universal skill that transcends any single product or vendor.

Wrapping Up

SuiteQL made us better NetSuite developers because it forced us to understand how the system thinks. Prompt engineering will do the same — just at a higher level of abstraction.

And while the tools will keep changing, the skill of clearly defining intent, context, and outcome will never go out of style.