The 5 Love Languages of Generative AI

Welcome & Introduction

Dean Peters | 00:00–00:07
I think we’re about ready to start here, don’t you?

Robyn Brooks | 00:07–00:10
Yes, let’s go ahead and kick off.

Robyn Brooks | 00:10–00:52
Welcome everyone to today’s webinar on The Five Love Languages of Generative AI. We’re going to have a good conversation about how to use generative AI tools to improve your product management game.

Let me tell you a little about the Productside. We are an outcome-driven product partner, offering training and consulting services to individuals and teams who are looking to improve their product management practice. We’re excited to share some information with you today.

A little bit about us: I’m Robyn Brooks, Director of Product Management here at Productside. I work on our curriculum and consulting products for product managers and product teams.

Dean Peters | 00:52–01:19
Thank you! And if you’re looking at your screen, yes — I’m the one with the face for radio. I’ll also be one of your hosts today. I’m a Principal Consultant and trainer. I see I’ve got some former students here who’ve attended my Agile Product Management, Optimal Product Management, and Digital Product Management classes. Glad to teach those.

If you’re not familiar with how we run these webinars, let’s go over a few rules of the road.

FAQ & Housekeeping

Dean Peters | 01:19–02:10
Will this be recorded?
Absolutely — yes, it will be recorded.
And in fact, if someone arrives late and asks in the chat, “Will it be recorded?” you have my permission to say, “Yes, it will be recorded.”

Will we be answering questions?
If you use the Q&A box — and if we have time and the question is relevant — we’ll answer it. If anything is missed, we’ll try to follow up afterwards.

Are we available for consulting, training, and optimization?
You betcha — and we’ll give you some links to contact us.

Tools you’ll see today:

• ChatGPT Plus
• Claude
• Bard
• Bing Chat
• Zoom
• Mural

We’ve already shared the Mural link with you.

Webinar Participation Guidelines

Dean Peters | 02:10–02:48
We want to keep this a safe and fun place.

So please:

• Be respectful of each other
• Participate, but stay on topic
• Don’t share anything private
• Use the Chat box for comments
• Use Q&A for questions
• Respect the time limits
• And above all — have fun

You’ll see a lot of navigation markers on the screen, but I’ll walk us through everything.

Poll #1 — How Useful Is Generative AI in Your Work?

Dean Peters | 02:48–03:17
Let’s run our first poll to get a feel for the room.

On a scale of 1 to 5, how useful is generative AI in your work as a product manager?
(If you’re not a PM, pretend you are.)

Options:

1. It rocks
2. It helps
3. It tries hard, but you have to fix it
4. It’s spam
5. It sucks — it’s dead to me
6. I don’t know about this GenAI thing — it’s dead to me

Dean Peters | 03:17–03:31
Give people a few more seconds…

Dean Peters | 03:31–03:45
All right — let’s end the poll and look at the results.

Dean Peters | 03:45–04:05
It looks like everyone’s mostly in the middle: “It tries hard, but close — no cigar.”
Thanks for participating!
So we’re going to talk about how to deal with that.

Setting Up the Conversation — The Joke About “Delivery”

Robyn Brooks | 04:05–04:10
Dean, I hear you have a joke for me?

Dean Peters | 04:10–05:13
Yes — and this one takes me back to when I was a summer camp counselor. I was there for five summers. By the fifth year, all of us veteran counselors had numbered our jokes, so instead of telling them, we’d just shout the number of the punchline.

But that last year, I got assigned a junior counselor — a high schooler named Trevor. Total newbie. I had to train him.

He hears us shouting numbers and laughing, and he asks, “Hey Dean, what’s the best joke to tell?”
I said, “Well, you’ve got to be careful — I like number 42, but—”
Before I can finish warning him, Trevor yells out:
“FORTY-TWO!”
He’s so proud of himself. But nobody laughs.

The director comes up to him and says, “Trevor, great joke — but it’s all about the delivery.”

And if that’s confusing, think about how confused ChatGPT or Claude is when you come up to that blank prompt window and just shout the punchline at it.

“Shouting the Punchline” — Bad Prompts

Dean Peters | 05:13–06:03
Imagine walking up to a generative AI tool and doing this:

“Give me a market analysis.”
“Pet care on demand.”
“Build me a product roadmap.”

It doesn’t know:

• Who you are
• Your role
• Your product
• Your customers
• Your context

And then you wonder why it feels off.

That’s because you’re shouting the punchline.

Robyn Brooks | 06:03–06:24
Yes — I love the “Tell me about a pet care app” example. The vaguer your prompt, the more likely the tool is to make something up. These tools have a reputation for hallucinating — but the less you give them, the more likely they are to wander.

Prompt Whispering — “What’s My Motivation?”

Dean Peters | 06:24–07:14
Exactly. You can’t shout at your generative AI — you need to be a prompt whisperer.

Since Robyn and I both have theatre backgrounds, let’s channel Stanislavski.

Imagine AI as an actor with no clue who they are. They stand on stage waiting.

They need:

• Backstory
• Motivation
• Stage directions
• Character relationships
• Context

When you give AI a vague one-liner, it’s like asking an actor to perform with no script.

Robyn Brooks | 07:14–07:31
Exactly. When we talk to generative AI, we need to give it enough context to perform the way we want.

Poll #2 — How Do You Give AI Context?

Dean Peters | 07:31–07:58
Let’s run another poll.

Which strategies are you using to give generative AI the context it needs?
(Check all that apply.)

• Tell it a story
• Provide lots of background text
• Use personas or segments
• Provide examples (few-shot prompting)
• Fine-tune with human feedback

Dean Peters | 07:58–08:23
Looks like we’re seeing a good mix. Some people are supplying background and examples, but fewer are telling the story upfront. We’ll talk about why that matters.

Because today is all about working more productively as product folks using generative AI.

The Five Love Languages of AI

Dean Peters | 08:23–08:57
Let’s get into the core of today’s theme: The Five Love Languages of Generative AI.

You may know the book “The Five Love Languages.” We borrowed the framework — not the content. The AI version is very different, but the metaphor works beautifully because we are essentially building a relationship with these tools.

Each “love language” is a different method to help AI understand what we want and produce better outcomes.

Here’s what they are:

1. Acts of Service — Giving AI a task
2. Words of Affirmation — Framing the response
3. Quality Time — Iterating through refinement
4. Receiving Gifts — Supplying examples (“few-shot prompting”)
5. Physical Touch — Structuring instructions, formats, constraints

We’ll go through each one with examples.

Love Language #1 — Acts of Service

Dean Peters | 08:57–09:48
The first love language is Acts of Service: You’re literally giving AI a job to do.

But here’s the thing — you have to tell it:

• What the task is
• Why you need it
• For whom
• In what tone
• In what format
• With what constraints

If you don’t, it defaults to “Wikipedia-plus-enthusiasm,” which is rarely what we need.

Use “Acts of Service” prompts like this:

“You are an experienced Product Manager working with a cross-functional team. Your task is to summarize this customer interview into three insights and three risks. Use bullet points. Keep it neutral and non-salesy.”

That’s a service-oriented prompt: clear, actionable, role-defined.

Love Language #2 — Words of Affirmation

Robyn Brooks | 09:48–10:12
This one is my favorite. It’s about framing tone and intent.

You are essentially “complimenting” the output you want.

Examples:

“Give me a concise, executive-ready summary.”
“Provide a friendly, human explanation for non-technical users.”
“Respond like a supportive coach helping a junior PM.”

These aren’t really compliments — they’re tonal guides. But AI responds to them extremely well.

Love Language #3 — Quality Time

Dean Peters | 10:12–11:08
Quality Time means: don’t expect perfect results in one shot.

Generative AI shines in iterative refinement.

Ask it:

• “Rewrite this.”
• “Shorten this.”
• “Make this more specific.”
• “Give me 3 variations.”
• “Now merge version #2 and #3.”

This iterative approach is where AI starts feeling like a partner instead of a vending machine.

Quality Time prompts usually begin with:

“Let’s refine this together…”
“Try another version…”
“Let’s iterate…”

When you collaborate, the output improves exponentially.

Love Language #4 — Receiving Gifts

Robyn Brooks | 11:08–11:36
The “gift” in this case is examples.

AI learns by imitation. So when you give it a sample, it will match patterns, structure, tone, and format.

Example prompt:

“Here is an example of the style I want. Rewrite my content in the same tone.”

This is called few-shot prompting.
And it massively increases accuracy.

Love Language #5 — Physical Touch

Dean Peters | 11:36–12:12
Don’t worry — nobody’s touching AI.

“Physical touch” here refers to structure:

• Templates
• Outlines
• Tables
• Headings
• Formats
• JSON
• Bullet lists

The more structure you give, the more accurate the output becomes.

AI loves structure. It removes ambiguity.
It is the single best way to reduce hallucinations.

Putting the Love Languages Together

Dean Peters | 12:12–12:47
When you combine these love languages — the results are dramatically better than any single prompting trick.

A great prompt will often use:

• A task
• A persona
• Tone/style framing
• Examples
• Formatting instructions
• Constraints
• And collaboration prompts

That combination makes AI feel like a teammate instead of a random intern with too much confidence.

Live Demo Setup

Robyn Brooks | 12:47–13:11
All right — now that we’ve laid out the framework, we’re going to move into some live demos using the mural board you all have access to.

We’ll show you how the love languages work in practice — and how different AI tools handle the same prompt.

Demo: How Different AI Tools Respond

Dean Peters | 13:11–13:54
Let’s kick off with a simple example:
We’ll give ChatGPT, Claude, and Bard the same prompt and observe how each interprets the instructions.

This is a great way to see:

• Which tool handles nuance better
• Which is more verbose
• Which stays closer to your structure
• Which requires more follow-up refinement

As you watch, ask yourself:
Which tool matches my cognitive style? Which one is easiest for me to work with?

Example Prompt Breakdown

Dean Peters | 13:54–14:43
Here’s the base prompt we’ll use:

“You are a senior product manager at a global SaaS company. Your task is to take this raw customer interview transcript and summarize it into:

1. Three key insights

2. Three risks

3. Two follow-up questionsMake the output concise and executive-ready.”

Notice what we included:

• Role
• Context
• Task
• Deliverables
• Formatting
• Tone

That’s at least four of the love languages baked into one prompt.

Tool Comparison Results

Dean Peters | 14:43–15:15
Now, in our tests:

• ChatGPT tends to be the most polished and structured
• Claude tends to be the best for long-form reasoning and accuracy
• Bard is surprisingly creative but sometimes drifts
• Bing is concise but can be shallow without guidance

This is why prompt structure matters.
It narrows variance and reduces hallucination.

Poll #3 — Which AI Tool Do You Use Most?

Robyn Brooks | 15:15–15:43
Let’s run a quick poll:

Which generative AI tool do you use most often?

• ChatGPT
• Claude
• Bard
• Bing
• Other
• I don’t use any yet

Robyn Brooks | 15:43–16:10
All right — votes are coming in…
ChatGPT is dominating, as expected. Claude is gaining ground though!

Demo: Turning “Bad Prompts” into “Good Prompts”

Dean Peters | 16:10–17:08
For the next demo, let’s take some bad prompts from the audience mural board and show how to improve them.

Example bad prompt:

“Help me with my roadmap.”

If you do that, AI is thinking:

“What roadmap? For whom? For what product? For what market? For what timeline? For what business goals? For what audience?”

Here’s the improved version:

“You are a senior product strategist creating a six-month roadmap for a B2B analytics platform used by enterprise data teams. Prioritize initiatives using themes rather than features. The roadmap should be aligned to improving retention and increasing expansion revenue. Provide the output in a Now / Next / Later format.”

That’s night-and-day difference.

Demo: Applying All Five Love Languages Together

Dean Peters | 17:08–17:52
What we’re going to do now is take a real product scenario and use all five AI love languages to produce something useful — a product requirements draft.

This is where you’ll see that generative AI isn’t magic… it’s collaborative.

We start with:

• Context
• Role
• Task
• Tone and constraints
• Structure
• Examples (if you have them)

When we layer these together, the prompt becomes a performance cue — and AI becomes an actor with a script to follow.

Live Prompt Example #1 — PRD Drafting

Robyn Brooks | 17:52–18:22
Here’s a real prompt a PM might need:

“You are a senior product manager preparing a PRD for an upcoming feature that enables personalized onboarding within a mobile wellness app. Your task is to create a PRD outline including problem statement, goals, success metrics, user stories, edge cases, constraints, and open questions. Use clear headings and bullets. Keep it concise and professional.”

Look at how many love languages are at play here.

Live Demo Results

Dean Peters | 18:22–19:03
Here’s what ChatGPT produced from that prompt:

• A structured PRD
• A surprisingly accurate problem statement
• Goals aligned to business outcomes
• User stories that weren’t absurd
• Acceptance criteria
• Assumptions and risks
• Follow-up questions we could take to design and engineering

This is why structure matters.
This is why tone matters.
This is why quality time matters.

You don’t get this from a one-sentence punchline prompt.

Demo: Comparing Refinements

Robyn Brooks | 19:03–19:37
Now watch what happens when we refine the prompt:

“Now rewrite the PRD for a non-technical executive audience. Shorten the user stories. Reduce jargon. Move ‘constraints’ to the appendix.”

And right away, the tool adjusts.

This is the power of iterative prompting — Quality Time.

When Generative AI Gets It Wrong

Dean Peters | 19:37–20:12
Let’s talk about something important:
AI will still get things wrong.

It’s not because the model is dumb — it’s because:

• The prompt lacked constraints
• The prompt mixed tones (executive vs. technical)
• You asked for too many outcomes at once
• It filled in gaps with assumptions
• It hallucinated due to missing structure

That’s why collaborating with the tool — instead of shouting instructions — produces better results.

Poll #4 — What’s Your Biggest AI Challenge?

Robyn Brooks | 20:12–20:45
Let’s do another poll.

What is your biggest challenge when using generative AI?

• Getting it to understand context
• Getting it to follow instructions
• Getting it to stop hallucinating
• Getting it to be concise
• Getting it to produce something usable in one try
• I don’t know what to ask it

Robyn Brooks | 20:45–21:05
All right — lots of responses.
The top issues are context and hallucination — very common.

Discussion: Why Hallucinations Happen

Dean Peters | 21:05–21:38
Hallucinations happen because:

• AI predicts the next most probable words
• It fills in missing details you didn’t specify
• It tilts toward confidence, even when wrong
• It doesn’t know your environment, tech stack, or users
• It tries to “help” rather than “verify”

You can dramatically reduce hallucinations by giving strong constraints or asking it to cite uncertainties.

Example: Anti-Hallucination Prompt

Robyn Brooks | 21:38–22:09
Here’s a great technique:

“When you are unsure, say ‘I don’t know.’ Do not make up facts. Ask clarifying questions. Do not fabricate statistics.”

The model will follow that.

If you don’t want hallucination?
Tell it — directly.

Demo: Using Guardrails

Dean Peters | 22:09–22:43
Here’s another guardrail technique:

“Limit assumptions. Ask me three clarifying questions before generating the response.”

This transforms the experience into…
a conversation,
not a single-shot prompt.

And the output becomes dramatically higher quality.

Demo: Generating a Competitive Analysis

Robyn Brooks | 22:43–23:15
Here’s another real PM task:
Competitive analysis.

Bad prompt:

“Compare us to our competitors.”

Better prompt using the love languages:

“You are a senior product strategist analyzing three competitors in the telehealth market. Provide a comparison table including target users, pricing model, differentiators, major strengths, and major risks. Keep it to one page. Focus on insights relevant to a PM preparing an executive briefing.”

This gets you a useful output — not a Wikipedia dump.

Demo Results

Dean Peters | 23:15–23:52
We ran that exact prompt earlier and here’s what we got:

• A clean table
• Clear differentiators
• Real market positioning
• Only minimal hallucination
• Executive-ready wording
• Solid insights we could use as a draft

Again — the prompt is the difference.

Exercise on the Mural Board

Robyn Brooks | 23:52–24:23
Now, take a look at the mural board and find the “Bad Prompt Surgery” area.

Pick a prompt that someone posted, and let’s rewrite it using all five love languages.
Then we’ll run it through one of the AI tools live.

This is one of the best ways to develop prompt discipline — learning to see patterns and rewrite them quickly.

Dean Reviews Audience Prompts

Dean Peters | 24:23–25:07
I’m seeing some great “before” prompts already:

• “Help me with pricing.”
• “Give me a Q4 roadmap.”
• “Write a customer journey.”

These all need role, context, tone, constraints, and structure.
We’ll fix a few right now.

Here’s one improved version:

“Act as an experienced pricing strategist. Develop three monetization approaches for a B2C wellness app targeting women ages 25–40. Include pros, cons, revenue potential, and operational considerations. Present in a 3-column table.”

That’s what we mean by “love languages.”

Final Demo — Turning Messy Notes into a Strategy Brief

Robyn Brooks | 25:07–25:42
Our final demo takes a messy set of meeting notes — the kind we all have — and transforms them into a strategy brief.

Raw notes example:

• “Customers don’t use onboarding”
• “Need upsell flow?”
• “Sales thinks trial is too short”
• “Mobile bugs?”

Now give that to AI with structure:

“Turn these raw notes into a concise strategy brief with sections for: problem summary, opportunities, risks, required research, and next steps.”

Watch how AI makes order from chaos.

Closing Poll — Will You Use the Love Languages?

Dean Peters | 25:42–26:09
Final poll:

How likely are you to use the Five Love Languages of AI after today?

• Absolutely
• Probably
• Maybe
• Probably not
• No
• I already use them

Dean Peters | 26:09–26:23
Looks like a lot of “Absolutely” and “Probably.”
That’s great to see!

Wrap-Up

Robyn Brooks | 26:23–26:52
Thank you all so much for joining us.
We’ll be sending out:

• The recording
• The slides
• The mural board
• Additional prompting templates

We hope this helps you get more value out of generative AI in your day-to-day PM work.

Goodbye & Final Thoughts

Dean Peters | 26:52–27:14
Remember — when it comes to AI:

It’s all about the delivery.
And the relationship you build with these tools can genuinely elevate your product practice.

Thanks for being here.
Have a great rest of your day!