Dave Killeen: [00:00:00] Hello, hello, Dave Killeen here, Field CPO at Pendo EMEA, and a very warm welcome back to The Vibe PM, episode four, and for those just joining, welcome. This is the podcast where I'll be doing the deep dives for you on various different AI tools, tricks, and shenanigans, so you don't have to. I'll be bringing you super short, practical demos all under 10 minutes each, every week or so to help us all dance more effectively with AI and have more fun and impact in our roles.

Alrighty, let's kick this off with a rhetorical question. How many times have you asked the LLM for help and gotten back something utterly useless and then just give it up? Too many times right? And here's the thing, it's not the AI's fault of course. It's us. It's our prompts. For the most part, we prompt pretty poorly.

And most of us just throw half vague requests at these incredibly powerful tools and then wonder, why are we getting garbage back? And as you'll see in this episode, the difference between a mediocre prompt and an excellent [00:01:00] one isn't just incremental, it's transformational. We're talking about the gap between an AI response that creates more work for you.

The desk headbanging and having to go back and forth with it in quiet rage, versus the one that delivers the strategic asset that you're after, that you can then immediately use to drive decision-making straight away. So today, I'm gonna show you a tool that solves this problem completely. It's called the Prompt Improver from Claude Console, the Anthropic team, and it takes our vague prompts and turns it into magical ones.

No prompt engineering experience required for any of what I'm about to show you. And once you start doing what I'm about to show you, there really is no going back. Let's dive in.

Okay let's be honest about the core problem here. Most of us interact with AI like a lazy line manager. We drop vague one-line requests on its desk.

"Hey, look at this, please, thank you", and then we bugger off. We then act surprised when the result isn't very useful. We all know that a great prompt requires a great brief, but crafting that perfect [00:02:00] brief every time is a pain in the backside, and that's where the Prompt Improver from Claude Console completely changes the game.

It acts like an expert assistant force, taking that vague thaw and turning it into crystal clear brief that the AI can work far better with and execute perfectly. So over the next few minutes, I'm gonna show you exactly how this works and you are gonna love it. Let's dive in.

Okay, so to make this real, we need something to critique. So I'm gonna create a deliberately average product strategy document from scratch, and I'm gonna use voice. If you haven't seen episode one, Speech to Text, really important. You'll save a ton of time. Absolutely foundational. I'll be pulling on this whole theme right throughout this entire series. Check it out. You'll save so much time back in your day.

So now I'm in Claude, as you can see, and now I'm gonna activate voice and I'm gonna send in the following, imagine you're the Chief Product Officer of claude.ai by Anthropic. I want you to write me a three-page product strategy document for Claude, but I deliberately want you to make this in terms of a score at a 10, where 10 is excellent.

Uh, give this to me as a six. Thank you. [00:03:00] Always say thank you. Anyway, in that goes and then we are just gonna fast forward and here we go. It's a classic, good enough strategy doc. It ticks the boxes, some basic competitive analysis, a few standard metrics down the bottom. It's basically the kind of document that looks good in the surface, but doesn't actually inspire anyone or drive real clarity.

This is our perfect test subject. Now for the fun part, we're on the dashboard of Claude Console. I'm gonna click on generate prompt and using my voice input using Wispr Flow from episode one. I'll do the following, give me a critique of this product strategy document. In that goes, I'm gonna tick this box, which says here the prompt will be used with models that have thinking enabled.

That basically means that the prompt that's generated for you in the next page will be something that will work brilliantly with those deep thinking models. So I'll click generate and we'll fast forward. Would you look at this? You're an experienced business strategy consultant. This is so much better, right?

You are tasked to provide a comprehensive critique of this document. You're gonna create a scratch pad space, so all your working is gonna be upfront and [00:04:00] clear to the end user. And you're gonna go through all of these different vectors, market analysis, value proposition, execution plans, success metrics, risk assessment, it's bonkers.

And then once you've gone through that as a scratch pad, then you're gonna take all of this into an executive summary. The key strengths, critical weaknesses, detailed analysis across all of what you see here. This is bonkers. Recommendations. The overall rating. Also, tell us how good do you think this actually is?

Your critique should be constructive, specific, actionable. I mean, honestly, war and peace, it's fantastic. That took no time at all. If you do this, honestly, it is a game-changer. You'll never go back. Anyway, let's move on to the next stage.

And now like a cookery program, we take them outta the oven. Here's both of them side by side.

So on the left hand side, you've got the very, very simple prompt and it's got this very, very basic critique of the strategy document. Now they copied and pasted above up here, and yeah, it's okay. It's fine. Um, you get some sense of where it's at here right now. [00:05:00] Now just quickly, just jump over onto the right-hand side.

Same document. But the larger, larger, larger prompt that was generated for us, by the way, in mere seconds is we upfront have then the scratch pad, right? So here's the scratch pad where it then generates its thinking, so it can refer to that as it proceeds with the task at hand. And then it just jumps into your executive summary, and it goes into an awful lot more detail as you can see here.

Asking for realistic timelines, and I'm just not gonna go through everything here, but you can see it all here. For those just listening, I'm sorry, but it's hard to go through all of this in the time that I have. But you can see it's war and peace of a response of a helpful response, a brilliantly helpful response, and then at the end, an overall rating, which actually interestingly says it's a six outta ten.

And funny enough, that is actually what we said we wanted in the original prompt. If you remember when we had this dummy product strategy created in the first place. So that's it. Just showing you side by side the difference that you get. And it is very, very, very different. Far, far richer, far more useful, more of a critique for you to [00:06:00] help.

So what's the secret sauce here? Why is this new prompt a total game-changer? Well, it's simple. You've stopped treating the AI like a clueless intern and started treating it like a highly priced expert that you've hired for the day. Think about it. First, you gave it a clear role. You are a top-tier strategy consultant, and you're not getting a generic opinion.

You're getting an expert's analysis, and next up, then you gave it a statement of work, right? You've defined the exact deliverables. I need you to assess the specific areas and structure your feedback with an exact summary and recommendations, et cetera, et cetera. You've removed all of that ambiguity that you had in that earlier crappier prompt.

And most importantly, you set the standard of quality, right? You've told it. Don't just give me fluffy observations. I expect you to think, to analyze, and to deliver insights that a real expert would, and that's how you get a document that drives clarity instead of just a list of comments. And so how do we plug all this into our workflow and start leveling up right now?

[00:07:00] This is how you go from being a good PM to being a great one. First, I think you should be de-risking your documents. Every single PRD, strategy doc, analysis that you write now gets reviewed by your on-demand AI consultant before anyone sees it. You'll catch 90% of the issues yourself making your work tighter, smarter, and way more credible from the get go.

Second, you win your arguments. Before you try and persuade anyone, especially your execs, you run your pitch through a brutal, improved critique. The AI will act as the ultimate devil's advocate for you, strengthening your narrative and making sure that your logic is undeniable. You'll start winning over stakeholders a lot more easily.

And third, elevate your team. Stop giving lazy feedback as a manager. Use this to provide your team with the kind of structured, insightful guidance that actually helps 'em grow.

So what's the cost of all of this magic? Well, today the cost is $3 for a million input tokens and $15 for a million output tokens.

The demo that I [00:08:00] showed you today is about 1000 tokens in total, in and out. So that is thousands of percent. So really for the price of a cup of coffee, you could be dancing with this, of course, the entire year, honestly nuts.

So here's your challenge. It's a simple one. Treat bad AI results as a bug that you need to fix.

Your mission is to take one, just one of your standard generic prompts this week, and run it through the prompt improver. See the transformation for yourself. I promise you, this is one way door decision. Once you start seeing the quality of the results you can get, you'll find it physically painful to go back to the old way.

And so that's it for this episode, and as always, we deliberately keep all of these episodes under 10 minutes for a reason, to give you powerful, practical tactics in snackable form that you can get more time back in your day to truly focus on what matters, thinking more strategically and spending more time with your customers.

So if this was valuable, please share it with your team because when everyone's [00:09:00] tools get sharper, you don't just work faster as individuals. Collectively, you can have a much bigger impact within your organization and honestly have a lot more fun in your job. Thank you for joining me on The Vibe PM. Please spread the vibes and I'll see you next time.