How to Get the Best Results from AI in 3 Steps

Let me let you in on a little secret…

The quality of your AI outputs depends entirely on the quality of your inputs.

It’s what we call “Prompt Engineering.”

Now, before you freak out at the word "engineering"—it’s not as technical as it sounds.

It’s nothing else than systematically providing the AI with the correct instructions such that it produces the right outputs.

It’s called “engineering” because it’s a repetitive process that improves the prompt by comparing the current output with previous outputs.

It gets there step-by-step.

Here’s a simple 3-step prompting method you can use to get better AI results today.

It’s called Role, Task, Format (RTF).

Here’s how it works:

Step 1: Role

You tell the AI what role it’s playing. Let’s say you want it to draft a cold outreach message.

For example:

You’re a proficient copywriter who specializes in writing cold outreach messages to new LinkedIn contacts

Step 2: Task

You give it a specific task.

For example:

Draft an introduction message for a contact I’ve recently added on LinkedIn.

Step 3: Format

You specify how you want the output structured.

In this case:

Your message must begin with a greeting, then include an impressive accomplishment the new contact listed on their profile, and lastly say that I’d like to know more about the projects the contact is currently working on. You must use clear language that’s easy to read. Your message isn’t allowed to contain any salesy language or posh words.

Boom.

Role. Task. Format.

That’s it!

Here is ChatGPT’s response:

Not bad! Can we personalize it more? 

The message is too generic. Unless you’re one of those shady copy-pasters, messaging you 3x per day about their awesome “piping hot” leads, you need something more relevant!

How about using your own message template? You want your brand voice to get through, after all. And you’re a human, not a machine (like the spammers?...)

Then step 3, Format, becomes:

Your message must adhere to the following format: “Hi [name], saw that you’re [condense their job title into one word] in [condense the industry they’re in into one word] in London. I work as a generative IA consultant, usually living in Barcelona, but staying in London for a few weeks to meet cool people. You caught my attention with [insert a short and impressive fact from their profile description]. Looking forward to keeping in touch! What are you currently working on that you’re excited about? Cheers, Robert”

Notice how we put placeholders directly in the message between brackets [ and ]?

Notice how the placeholders contain mini-instructions specifying what information we want and how it must be formatted?

The AI is so smart that it recognizes what must be replaced and what must be kept the same!

In my whole career as a software engineer, I’ve never used placeholders so easily! Believe me, this technology is a game-changer.

Let’s keep going—we’re only halfway there.

Where does the information come from that the AI uses to fill in the placeholders?

It’s simple: We just provide it!

Step 2, Task, then becomes:

You’ll be using the following person’s LinkedIn profile information:

- Name: "Robert Hangu
- LinkedIn headline: "Gen AI Consultant and Product Builder [...]"
- Additional information: "Founder Next Jobs. Let AI match you with jobs that suit your career credentials, personality, and ideal work environment! [...]".
- LinkedIn profile description (About section): "I'm a generative AI product builder and consultant. My products and solutions use generative AI to empower end users. [...]”

Curious to see the result? Here it is:

😮

If your jaw hasn’t dropped by now, I’m sure the “piping hot” spammer’s definitely has!

Can you see how it precisely replaced the placeholders and nothing more(!) with a smartly condensed text?

It inserted the name “Robert”, the job title “consultant” and the industry “AI”. Then it provided an impressive fact from their profile “you vision to make work obsolete [...]”.

Notice how it kept everything else untouched, preserving the rest of our original message?

Wow, now that’s a handy tool to have in your arsenal!

And here’s the best part…

Once you get the hang of this framework, you’ll start seeing more accurate and polished outputs from your AI.

So if you’re tired of getting generic, so-so responses, try this RTF method on your next prompt.

And hey, if you’re currently in a situation where this could help you, feel free to ask me your question. Just reply to this email and ask it right away.

Or if you have a bigger project you want to discuss, you can book a free discovery call with me, by clicking the link below:

Book a Free Discovery Call

Cheers,
Robert