[On Demand] Product Management Webinar: AI in Product Management

Megan Saker: [00:00:00] Hello everyone. Welcome to ProdPad‘s webinar today. So we are AI prompts for product teams on how to harness the power of product management AI so just before we kick off i’ll just quickly introduce myself.

I’m Megan. I’m the CMO here at ProdPad and for those of you who maybe don’t know who ProdPad are. We are a product management software company. So we are a complete product management platform that is intended to give product managers and product teams a home for your strategy and your planning.

So you’re able to house your roadmapping, your idea management and your customer feedback management all together in one place to provide that single source of truth where you can capture your strategy, capture your ideas, work ideas through your workflow, both then that’s pre development even pre roadmap.

And to help you communicate your roadmap to your stakeholders, to your customers, there’s roadmap publishing options in there as well. But not only is ProdPad about roadmapping, idea management and customer feedback, but we’ve also got a pretty good AI. Since we’re here today to talk about AI and product management, let me just briefly mention CoPilot.

So for those of you who haven’t used CoPilot or come across CoPilot yet, CoPilot is the AI built into ProdPad. It’s by far the most relevant application of AI for product management for the product management context, but we will talk about

Before I introduce Janna and we kick off the webinar properly, let me just do a bit of housekeeping. So we will make sure that we leave some time at the end. We’ll try our best to leave some time at the end for a Q&A. If you look at the bottom of your Zoom window here, you’ll find a specific box, a specific [00:02:00] icon for a Q& A.

So that’s where to put your questions ready for the Q& A at the end. Add them whenever you think of them during the webinar and Jen and I will come to them at the end. We also have the chat function, which obviously we’re using now and everyone’s using there. Feel free to use that all throughout the webinar.

That’s an open space. for us to all talk about the topics that Janna’s covering to, yeah, network with your peers. So please use that. Also in the chat, you’ll find two of our colleagues, Janna and I’s colleagues. We’ve got Simon, who is Janna’s co-founder and our CTO. So he’s knocking around in the chat, and knows an awful lot about AI.

And we’ve also got Richard. Who is one of our product experts and he is around to answer any questions. And if you’ve got any questions specifically about ProdPad, then Rich is your guy. And he’s always around if you want to have a call or anything about anything in more detail later on.

That is the, oh, the final [00:03:00] thing to say housekeeping wise is that we will send an email after the webinar, just give us a day or two, we’ll send you an email with any links that we talk about, any extra resources, but also the recording of the webinar today. Let me stop waffling on because you’re not here to listen to me.

Let me introduce Janna. So hopefully most of Janna Bastow. Janna is Podpad’s CEO and co-founder. She’s the inventor of the Now, Next, Later roadmap, and she’s pretty much an authority on AI in the product management context. So without further ado, let me hand over to Janna. 

Janna Bastow: Excellent. Hey, thank you so much for that warm intro.

Great to see everybody here. I see the chat lighting up as well. Megan’s going to be on hand to help keep an eye on that as we go through. But today we are going to be talking about product management, AI, as well as having a chance to talk about that CoPilot tool. I’ll have a chance to show you through.

So that’s going to be very cool. So before we dive in, let’s set the [00:04:00] stage for today’s discussion on AI and product management. So there’s two key facets of this. One is how you as a product manager can use AI to make your day to day work more effective and efficient, and how you can integrate AI into your product to gain a competitive edge and solve bigger, better problems.

So when it comes to that, there’s a few different approaches. One using an existing AI model to enhance your current product. One is building your own AI model within an existing product. And the third one is managing an AI first product that’s built from the ground up with AI at its core.

But today we’re going to be diving into that first area. Using AI as a product manager to boost your productivity. If you want to know more about that second area, adding AI capabilities into your own product, we have a definitive ebook that’s on its way out that’s going to cover everything.

It was just too much to cover in one day. If you can’t wait for the e-book, we have a couple [00:05:00] articles written by our CTO and my co-founder Simon Cast. He’s just been saying hello in the chat. Hi Simon, thanks for joining today. There’s a couple QR codes that you can scan that will take you here. We’ll also include these links in the email that we’ll send out afterwards with the recording of this webinar.

So today we’re going to talk about first we’re going to be diving into, let’s talk about the buzz around generative AI, there’s this mix of fear and excitement that’s been generated around it. We want to talk about why it’s important to address this up front. Next,

Before we dive into how to prompt AI, I want to outline exactly what AI can do for product managers, specifically which PM tasks it can help you do faster or better. And then we’ll get into the core session the principles of prompt engineering and how to craft effective prompts to get PM specific results.

I’ve got examples that I can share that are going to guide us. And then I want to explore an alternative approach that reduces the need for heavy prompting. And finally, to make it more practical, we’ll dive, we’ll jump in with a live [00:06:00] demo of prompts in action with our ProdPad CoPilot. So are we ready to jump in?

Is AI a good thing or a bad thing? Let’s be real. If you’re here learning how to prompt AI effectively, you’re probably leaning towards the good side rather than running for cover. But it’s worth addressing the big question. Will AI replace product managers? So my take on that is no, actually quite the opposite.

AI is here to make us better, more valuable, and push the product management discipline further. We’ve already seen this field evolve from the project management days of old to agile, lean, and strategic roadmapping that we’re doing today. AI can actually help accelerate the progression. It frees us up from grunt work, allowing us to focus on high value strategic decision making.

And when we spend more time delivering outcomes that matter, we solidify our role as key strategic, as a key strategic function within our companies. And no, AI is not some RoboCop style threat that’s going to burst in and annihilate us all. We don’t have to worry about that. Actually think about AI as a [00:07:00] tool, right?

Remember when Photoshop was launched? Like photographers didn’t vanish. They leaned into it. They learned how to use it and their creative potential skyrocketed. Those who mastered the new tech unlocked bigger possibilities. So AI is like a new Photoshop, right? We’re here to learn to use it and start unlocking new levels of impact in our roles as product people.

So when you learn to use AI effectively, the benefits are huge, right? That’s why prompt engineering is a skill worth having. Like here’s what AI can do for you. It’s going to help you save time. Like with the right prompt AI can generate highly relevant, well written outputs faster than you could draft them yourself.

Whether you’re summarizing or starting from scratch. It can also be your brainstorming sidekick, offering fresh ideas on product differentiation and problem solving or strategy with just one prompt. And it can speed up iteration, from concept testing to prototyping and MVPs, AI can help you iterate, refine and [00:08:00] adapt faster by automating tasks like writing code for a usable proof of concept.

And it doesn’t just help you, it helps your entire product team by automating repetitive tasks and freeing up time for high impact work. And, it’s also not just a writing tool. A lot of people just use it for writing, but it can offer guidance, it can suggest best practices, it can answer questions.

Especially when trained on PM specific knowledge, like our ProdPad CoPilot. So right now, being skilled with AI can give you a competitive edge in the job market. But soon, prompt engineering is going to be a core skill. Everyone’s going to know how to use it. That’s going to be just a basic expectation.

Now the good news is that you’re here today. You’re ahead of the curve already. So I want to dive into how to craft effective AI prompts to get you started. What exactly can it help you with as a product manager? Where can it deliver on those time savings and performance boosts that I’ve promised?

So with the right prompts, AI can support pretty much any stage of product development from [00:09:00] discovery and prioritization to documentation, stakeholder management and beyond. So let’s break down some of the key areas where AI is going to make a difference, starting with product strategy. So generative AI can really help you craft and communicate your overall strategy.

We all know how important it is to clearly express your ambitions and priorities. Without that clarity, you won’t get alignment across teams. A strong strategy needs to remove ambiguity and inspire your team to rally around the vision. A great place to start with AI is by working on your vision statement.

So you can describe your product and what you want to achieve and AI will generate a motivating, clear vision statement for you that you can build on. Or you can share your existing statement and ask AI for constructive feedback and suggestions on how to improve it. So AI is particularly helpful if you tend to think in short form, like bullet points, you need to convert that into, more creative, polished.

But also it can work the opposite. If you tend [00:10:00] to write way too much, it can help shrink it down into something that is more digestible by your audience. So with CoPilot and ProdPad, it’s even easier. You don’t have to prompt it at all. You just enter your draft vision statement into the strategy canvas.

Click a button and CoPilot provides feedback and suggestions instantly. Really helpful for that. And I also want to talk about goal setting. So AI can help you define your objectives and key results. With tools like ChatGPT, you need to provide a lot of context, your product details, your vision, your overall strategy.

But once you give that information, you can ask AI to generate relevant goals and objectives. Be sure to specify your preferred framework, like OKRs, and the format that you want the output in. CoPilot makes this a lot simpler. In ProdPad, you can enter your objective and CoPilot will automatically generate specific measurable key results with a click.

No need for extra prompting or setup, and you know that what it’s creating are [00:11:00] outcome focused leading key measures for your key results. And finally, AI can help with ideation, sparking ideas and kick starting your creative thinking. You can use AI to generate ideas for a whole bunch of strategic needs, like new product ideas to address market problems that you’re seeing out there, new roadmap initiatives that align with your vision and your objectives feature ideas that fit within broader roadmap initiatives.

Again, CoPilot and ProdPads streamlines this process. You can generate initiatives and ideas with a single click. No manual prompts, no typing required. Or if you want to brainstorm interactively, then you can use the chat interface to bounce ideas around with CoPilot. So I use this feature all the time.

It’s one of my favorite things now, especially when I need to tackle a problem from a fresh perspective. So these AI capabilities can save you time and help you focus on high level strategic work, pushing your product management efforts to the next level. And I also want to talk about [00:12:00] discovery. And where can I help here?

But I want to start with a bit of caution, like AI can definitely support your research efforts, but, only in certain ways, when it comes to market or competitor analysis, you need up to date data, right? The problem is that some of these AI models have a knowledge cutoff.

Like for instance GPT 4’s cutoff is October 2023, I think. So if you ask it for the latest feature comparison grid between you and your competitor, it might not have that information. It might not be accurate. So you’ve got to use it with some care. But that doesn’t mean that AI can’t help, like where it can be useful, you can ask it for a general assessment of a market to get a solid starting point, or you can ask it to fact check and refine and do research for you.

If you’re stuck on structuring a market analysis report, it can create that rough outline to get you started, get rid of that dreaded blank page. It can also be used to digest and summarize recent industry reports or your [00:13:00] competitor’s annual reports that they put out there.

And basically have it tell you what the page is about, what the paper is about. Saves you a ton of reading. And you know what? The same goes for user research. AI can help, but you need to know its limits. Talking to real or potential customers should always be your main source of insight.

Not going to replace that, but AI can still lighten your load, right? So you can use it to suggest research methodologies or generate interview or focus group questions or write tests or, form for form fields or test scripts for user testing. You can help it. You can use it to help you prepare research reports and presentations and very importantly, you can have it analyze the data from your research to help you spot trends and draw conclusions faster.

What AI can’t do is replace real validation. So don’t ask it things like, would a customer of a mobile banking app find a budgeting tool useful and then trust the answer blindly. So AI can guide you, but the [00:14:00] real insights come from real people. And talking about data analysis, this is where I can be a real saver, right?

So it can analyze large volumes of data like customer feedback or usage patterns and quickly surface common themes, patterns or anomalies. So gone are the days of running hours, long hours, long affinity mapping sessions to spot trends. manually. That said, be mindful of the tool that you’re using. So with generated, with general AI tools you’ll need to export and format your data before uploading it, loading it.

And you’ll need to explain the context in your prompt so that AI knows what it’s looking at. It’s doing, but it’s a bit of a hassle. A better option is to use a data capture tool with built in AI capabilities tools. already have access to your data, so there’s no exporting, importing, or extra explaining involved.

So ProdPad’s AI signals tool is a really good example of this. It automatically scans your customer feedback and then [00:15:00] surfaces the signals, the key themes, without you having to lift a finger, really. And then with CoPilot, paired with that, you can ask any question about your feedback and get an instant actionable answer.

So CoPilot also links feedback directly to relevant ideas in your backlog, so you don’t have to manually connect the dots. And then also, let’s talk about prototyping. AI can help you get a prototype off the ground really quickly, even if you don’t have immediate access to development resources. There’s specialist AI coding tools, like Cursor or others, that can write production ready code.

But AI models like ChatGP, ChatGPT can handle some of this too. You just need to give them well crafted prompts and feed in designs or That is your PRD. And from there, AI can generate some of the code that you need to bring that prototype to life. So whether you’re building a, um, quick proof of concept or iterating ideas this is a few [00:16:00] ways that AI can help you move faster and have less friction.

And also let’s talk about feedback management. So capturing feedback is the first step here. And AI can turn every customer interaction into a usable written record. Make sure everyone speaking to customers, whether on Zoom or Google Meet or any other tool, is using AI generated transcripts to capture what’s being said.

Or go one step further with AI note takers like Fathom, or there’s a bunch of others out there as well, Otter and other ones that I can’t remember. They don’t just describe. Here at ProgPad, our CSM team has made this even simpler by hooking up an integration that automatically pushes Fathom’s customer call summaries into our feedback system.

So we’ve cut out any of that manual uploading. There’s no missed insights. Everything is in there to be summarized, which is the next stage, right? You want to summarize feedback with AI. It’s a huge time [00:17:00] saver. So at ProdPad, our support tickets flow directly into the feedback system where there’s plenty of gems in there, but they can get pretty messy.

When you’ve got a support ticket with, the entire email chains or signatures clogging everything up. Again, CoPilot comes in handy here because every feedback entry has a button that you can click to instantly generate a clean summary complete with bullet points of the sentiment analysis and the key takeaways, and it just saves you from digging through cluttered messages to find the good stuff.

And then, analyzing feedback is where AI really shines, right? So with general AI tools, you have to export your feedback as a CSV, format it, and then explain what the data means which is actually a lot of hassle. But Signals and CoPilot and ProdPad, the AI has already got access to your feedback and can dive straight into the analysis.

It’s literally a one click, just analyze this feedback. And it identifies patterns, surfaces, themes, and even links directly to the relevant ideas in your backlog completely automatically. There’s no exporting, there’s no extra steps, just [00:18:00] actionable insights ready to use.

A few other places where it can help you out. Backlog management. The key here is having the right tool with AI capabilities built in. The AI should have access to what’s going on in your backlog, from the oldest ideas to the latest entries, and be able to connect everything to, to the insights that are feeding your prioritization process, like your feedback and your product vision.

All that grunt work of, Tagging, triaging, organizing, this is where AI can seriously help. In ProdPad, for example, we’ve got it set up to have duplicate ideas automatically flagged and removed. Feedback is linked to relevant ideas, so you don’t have to hunt for connections. Remember what was in your backlog.

You can also assess how well an idea aligns with your product vision with a simple click. This is one of my favourite because it acts as a coach for you for whether you’re on track or not. It provides reasons for the alignment or the lack thereof and then suggests improvements so that you can further improve that idea or iterate on [00:19:00] the direction that you’re taking.

Product documentation. It’s another big use case, so it can handle things like writing idea descriptions and drafting product requirement docs or specs creating user stories and acceptance criteria writing up release notes, whether they’re for internal teams or in product announcements, or, customer emails that you’re sending out there in prod pad, you don’t even have to craft these prompts.

We’ve got buttons throughout the app that allow you to just do things like generate user stories or, ask it to generate an idea description or just help you move faster with this stuff. And then finally in product copy is another area where AI can make. Your life is a lot easier, right?

So have it generate and suggest copy for things like empty states, tooltips, and in product messages, onboarding flows and product tours, upgrade paths, and conversion copy for upsells, and of course, as I mentioned, release announcements and things like that.

So [00:20:00] it can also help with your stakeholder comms. This is an area where time sinks happen all too often. We’ve talked about using AI for things like writing reports and communicating your strategy, but the right tool can take this even further and save you from all those interruptions and manual updates.

So here at ProdPad, we’ve always focused on making stakeholder communication teams with features like customizable roadmap views and external roadmap, publishing, automatic update notifications and really tight integrations of stuff like slack and teams. But now with CoPilot, it’s, we’ve taken it to another level, right?

So CoPilot. Is embedded within your product management system, meaning it has access to your roadmap, your backlog, your customer feedback, wait, what have you put out there for your strategy? What have you got for your OKRs and, everything else, all the decisions your team has made, it means it’s a game changer for handling the, all those day to day impromptu questions that you get from your stakeholders, your wider team and everybody else around you, common scenario, your boss [00:21:00] asks what’s on the roadmap that ties together to this strategic objective.

And yeah, they could check the roadmap themselves, and in ProdPad they could even group it by objective and get that information pretty easily, but let’s be honest, even with all that, they’re probably just going to come and ask you instead. With CoPilot, they don’t have to. They can just ask the CoPilot. It can tell them exactly which initiatives and ideas align with that objective, and provide links to each one and more info and tell them how it’s progressing and all sorts of stuff.

So the result is that stakeholders get the information they need without pulling you away from your deep focus work. Less time spent answering questions means more time to focus on the things that drive real product progress. And where you might have once gone digging through forums, asking online communities or endlessly searching for advice, you can now turn to AI for fast, reliable guidance on best practices within product management.

Sense checking the way you approach a certain product management job or asking for advice on the best way to do something is a great idea if you want to [00:22:00] be the best product manager you can be. So I always advocate the use of AI tools as sounding boards or as sidekicks as a coach to help you understand the best practice ways of working.

But of course the advice AI will give is only going to be as good as the advice that the model has been fed and trained with. Our co pilot has been built specifically with product management best practices in mind. It’s been fed with certain curated sources of best practice information to make sure it always delivers the best coaching and advice when you’re asking for it.

I want to drop a poll for you all. Let me pull this up for you. The first question I want to ask is what you’re using for AI help right now. I’m sorry. Let me just get this one up. Product management, AI poll. That’s the one. That’s the one launched. All right. You should have all been able to see that now.

So what are you using for AI at the moment? And I want to understand where you’d like to get AI help. What jobs would you like to farm out to that generative AI [00:23:00] or make faster with an AI assistant? If you have other jobs that aren’t on the list. Please share them in the chat but we’d love to see what’s coming up here, and I can see the answers coming in now, which is great.

Strategy documentation is, ooh, head to head with ideation very cool. I can see some other options coming up here. Marketing competitor analysis is really popular. Data analysis, no surprise there. Mark, user research planning. Sorry, 

Megan Saker: just to point out, make sure you can scroll in that window, so the second question.

There’s more options if you scroll down and the second question is hidden underneath there. 

Janna Bastow: Yeah, so absolutely feel free to pick from that big list here, there’s a lot of different options here. And AI for. We can see that we’ve got Ooh, lots and lots of people answering here. We’ve got 200 answers, which is great.

Getting some real insights here. Nice. 

Megan Saker: It’d be interesting to see the difference between what you’re using [00:24:00] AI for right now. Yeah. Versus, those jobs where you’re not using it yet but you want to. 

Janna Bastow: I’m seeing a discrepancy already, which is AI prototyping is something that people aren’t using for right now, but that they’d like to be able to.

So that’s very cool. That’s very cool. 

Megan Saker: Oh, maybe we do a separate webinar on that. Really drill into that. That’d be interesting. 

Janna Bastow: Yeah. I would love to talk about our process for AI prototyping. Backlog management is a really popular one. People aren’t using it for now, but they would really love to, which is fantastic because that’s something that I can show you today how that works.

Yeah. We can definitely help with that. Excellent. All right. So we feel like we’ve got enough answers. I’m going to end this here. Everyone’s had a chance to get their word in there. There we go. All right. Cool. So thanks for those insights. All so I want to talk about the best oh, I can share these [00:25:00] results.

There we go. It’s just popped up now. There we go. So as you guys are taking a look over those results, I want to get into how to get the best results when it comes to these tasks. I’ve shown you the potential, all the different ways that Gen AI can help you do more and move faster across the product management lifecycle.

If you’ve tried any of those AI tools, you know that you won’t necessarily get results that you’re happy with right off the bat. Certainly not with the most common generalist AI tools like chat GBT. So let me show you how to mask, how we can master prompt writing specifically for product management.

It’s going to make your life with this so much easier. Some general principles you need to remember when you’re engineering your prompts is to provide context. AI can’t read your mind. It needs relevant details to work effectively. So always include background information in your prompt where you have it.

If you’re asking for suggestions about your product, it needs to know what your product is. Feed it clear personal product goals, and the value prop. [00:26:00] Other information like that also really makes sure to keep it simple and structured, overloading the AI with too much detail can confuse it.

So focus on concise, goal oriented prompts for complex tasks, break them into smaller, more manageable parts. I’ll show you how to do that in a bit and use natural language. Just talk to it. AI responds to the best prompts written in everyday language, just like talking to a teammate. So don’t use weird robotic phrasing or an overly formal tone, just stick to clear conversational language to get the best results.

So I want to dive deeper on this. I want to introduce you to a really useful framework, the WISER. This was a framework created by Ali K. Miller, a well known voice in AI for Business, and it’s designed to give or help you give AI the right context and structure to generate top notch results. So let’s break it down.

what each letter stands for. W is who is it, right? Start by assigning the AI a role. Give it some context about who it’s supposed to be. Like, for [00:27:00] example, you are a product manager managing a B2B fintech SaaS platform. Why does this matter? Because when the AI knows who it’s supposed to be for, then sorry, who it’s supposed to be, then the response is more focused and relevant.

It is for instructions. So be specific about the task that you’re asking the AI to complete. For example, draft a high level GTM plan for a new expense tracking feature with key action points. Vague instructions will lead to vague results, so always be clear about what you want. The S stands for sub tasks, so break the task into smaller, manageable pieces to guide the AI step by step.

For example start by outlining the target audience, and then list three marketing channels, and then finally suggest KPIs to track success. This is especially important for complex tasks. Smaller steps lead to better, more actionable responses. E stands for examples. So provide a reference or example to help the AI understand what you’re looking for.

So an example here [00:28:00] is an example of a GTM plan we’ve used before. Align your responses with this structure, right? Show it what good looks like. It thrives on examples. So if you can give it a template or your previous work to model on, then you can get more on target results and review, right?

Don’t stop at the first output, refine and iterate by asking for adjustments. For example, add more detail to the target audience section or reformat this as a presentation outline. The first response isn’t always going to be perfect, but small tweaks can get you exactly where you need. So that’s the WISER framework.

Who, Instructions, Subtasks, Examples, and Review. I want to dive into each of them and talk about how you can use it for product management tasks.

So let’s break down the W step. Who is it? This is all about setting the right context. And it’s about more than just telling the AI who it is supposed to be, but you need to also include what they’re doing and why they’re [00:29:00] doing it. This combination sets the stage for more accurate and relevant responses.

For example, if we’re writing a prompt for product managers, the who is a product manager, the what is designing a mobile banking app, and the why is designed to help young people better manage their finances. There’s a couple other advanced prompting techniques that can help with this.

Domain priming involves instructing AI to adopt a specific role or perspective. You’re trying to make sure that it answers from a product management lens. You’re essentially priming it to think like a PM instead of, say, a general tech advisor or a marketing expert.

And then role playing, it, takes domain priming a step further. It’s a bit more creative. You can ask the AI to pretend to be a specific type of user or stakeholder to explore different perspectives. For example, if you’re brainstorming pain points for a particular user group, you can have the AI role play as a user and create

So let’s apply this to a real prompt by crafting the first section of it. So you are a product manager for a mobile [00:30:00] banking app. The app is designed specifically for young people aged 16 to 25 to help them develop financial acumen and better manage their finances. So that’s our starting point. I want to move to the next step, which is instructions where we’re going to clarify to the AI, what it is that we actually want them to do.

So this is where you can. tell the AI exactly what you need to deliver the instructions. If your instructions aren’t clear, the AI will make assumptions, which might leave you with results that don’t quite hit the mark. This is why precision is key when crafting this part of your prompt. So one prompting technique that you can use here is chain of thought.

It guides the AI to reason step by step, making it particularly useful for more complex tasks. For example, list three common objections personal banking customers might have to that might have, that they might have to use a budgeting feature. Then for each objection, suggest a solution or feature improvement to address it.

So you’re encouraging [00:31:00] well organized and actionable responses. So we can continue building our prompt here. The I section might look like something like this. Usage rates are low for our budgeting feature. We need ideas to solve this problem. List three common objections our customers might have to using the budgeting feature.

Then for each objection, suggest a solution or feature improvement to adjust it to address it. Present your ideas in a concise bullet point format, including how each solves the problem. So you’re taking a complex. process, and you’re breaking it down to smaller, more manageable tasks. Which takes us to the next step, which is subtasks.

Say you want your AI to map out how to increase feature adoption. This isn’t a one step task. It involves several stages. Understanding current usage, identifying what’s stopping users, finding friction points, brainstorming solutions, evaluating the feasibility, and we could go on. If you want all of that sorry, if you ask for all of that at once, you’re probably going to get a messy [00:32:00] response that doesn’t give you much to work with.

But if you break it down to smaller, more manageable tasks, subtasks, you’re going to get further. One great technique for this is prompt chaining. It involves connecting multiple prompts together, where each new prompt builds on the results of the previous one. It allows you to guide the AI step by step instead of overwhelming it with too much at once.

An example of this is, start by asking. List the common reasons why users don’t adopt new budgeting tools. And then once you have that, generate possible solutions for each identified reason, and then prompt it to assess something like, evaluate each solution for feasibility based on ease of implementation and potential impact.

So this structured approach keeps the AI focused and ensures that you get well thought out responses. One of the main reasons why prompts fail is if you’re asking for too much at once. So to continue on our example, the prompt might look something like this, after generating three solutions, rank them on these two criteria, technical [00:33:00] feasibility, as in how easy is it to implement each solution and impact and effort?

How effective will the solution be compared to the time and cost required? So it’s really Key to break it down into actionable steps. So you’re not getting this, jumbled response. And then next up is to provide an example of what you’re looking for. The example acts as a clear target.

It shows them what good looks like. It guides the AI’s reasoning and structure, so you’re more likely to get a result that matches your expectations. Plus by showing the AI what you’ve already considered, you reduce the risk of it repeating ideas that you’ve already explored. Great. Advanced technique for this is a few shot prompting.

This gives the AI a few examples of the type of response that you want, and then AI picks up on the patterns and generates something in line with those examples. So in the case of our example prompt, we could give the AI something like some existing ideas, like idea one. Implement an in app tutorial that explains how to use the budgeting [00:34:00] feature.

And idea two, gamify the budgeting feature by offering personalized incentives for people who complete goals when using it. So they set the bar for what AI should deliver. Now the last piece is to review and reflect, right? Really the last thing you want to do is review what the AI has given you.

And a really useful prompting technique here is reflection. It’s all about getting the AI to evaluate its own work, assessing its decisions, and identifying areas for improvement. So instead of taking the first response as is, you can guide the AI to critically evaluate and reflect on what it’s produced and then refine it further.

You’ve got to be clear about what you want the AI to review. It can base, it can vary based on your goals, but you’ll generally want it to reflect on clarity. Is it easy to understand and follow? Creativity, does it offer fresh, innovative ideas? Feasibility, are the proposed solutions feasible, and any gaps, did it miss anything important?

Our example prompt [00:35:00] might be something like, Review the proposed solutions, paying close attention to clarity, creativity, and feasibility. Identify any gaps in the response, or areas that were overlooked. Could certain aspects of the solutions better address the target audience’s pain points? Finally, reflect on whether the solutions could be made more actionable or user friendly.

So it really helps you to fine tune the outfit and you end up with the output and you end up with something much closer to what you really need. So there it is a well considered, crafted, prompt, built step by step up using the frame, the wiser framework now. Obviously, you don’t always need this to be you don’t always have to be this meticulous, right?

You can shoot from the hip a bit, you can fire off something short, or see what you get, and then refine as you go. And that’s totally fine for quick, low stakes tasks, but when it comes to tasks that you do a lot, like brainstorming feature ideas, and writing specs, and creating documentation, it’s worth taking the time to nail down a great front prompt up [00:36:00] front.

And then what you end up with is basically this reusable template. Next time the tasks come up. task comes up. You can just tweak the details and be off to the races. So that said, getting the perfect prompt can take some upfront effort. And yes, I know it might seem like a lot if you can’t be bothered with all that prompt crafting and to get the relevant output that’s from a product perspective, there is another option, right?

And let me show you what the alternative looks like. So on the left, you’ll see a prompt design for a general AI tool, like Chat GPT, built using the prompt engineering we just covered, carefully crafted step by step. And on the right, you’ll see the equivalent prompt you’d give CoPilot and ProdPad.

It’s much simpler. It already knows your context. You don’t need to provide all that extra background. You just give it a task and it’ll deliver the same or even better output with less and with less effort and faster less time to create all this stuff. So why can CoPilot [00:37:00] give you hyper relevant responses without needing all that?

It’s because ProdPad is a complete product management platform. It’s the home for your vision, your strategy, your roadmaps, your idea backlog, your feedback, your documentation, all the decisions that your team has made around your product. Everything your product team

And so we’ve developed this AI layer that sits right in the middle of it, and, CoPilot has access to everything. So it cuts out the need for that first step of prompt engineering, the who and the what and the why, because CoPilot already knows. It knows what your priorities are. It knows what your goals are.

But it also doesn’t stop there. It’s primed for a product perspective. Simon and our team have spent the last couple of years painstakingly feeding it some system instructions and refining its outputs. They’ve taught it to think how product teams think. Think about what they need to achieve and how they make decisions.

So there’s no need for you to manually prime it with prompts. CoPilot already acts like a product player on your team. And also it’s been [00:38:00] trained using curated sources of product management. Best practices. And the keyword here is curated. We’ve hand picked reliable, trustworthy sources that ensure that CoPilot gives you advice that you can trust based on quality information.

So it doesn’t just generate responses, it delivers practical, product specific advice that’s relevant and easy to use.

With all that, I’m sure some of you might want to see this in action. Is that about right? Let’s do it. Alright, excellent. 

Megan Saker: Oh good, there’s some eyes. Yes, people want to see it. 

Janna Bastow: Which 

Megan Saker: is 

Janna Bastow: excellent. People do want to see it. Excellent. Just pull up my demo account here and I’ll switch screen sharing and we’ll kick off.

Does that sound good? Nice. 

Megan Saker: Great. Let’s do it. All right. 

Let me share my screen. I think we’ve talked, you mentioned actually, a couple of places about the promptless. Instances of CoPilot. So those times where you can use [00:39:00] CoPilot and you don’t need, without the chat interface, without any need for prompting where you just click a button and the CoPilot will perform the task.

So maybe we start there. Maybe we can look at a few of those. Yeah, what about starting on a road map? Yeah, because you can generate a button for generating ideas, right? Absolutely. 

Janna Bastow: So within ProdPad, you’re able to manage as many products as you like, and then each one has its own roadmap.

But it’s also important to point out that it’s primed with information from your canvas, which we’ll come back to in a second. But basically, here’s my roadmap in ProdPad, and I can actually use it to outline where we are now, where we’re going down the line with as much or as little detail as needed.

So the ones that are later on tend to have less detail, the ones that are in the now tend to have more detail. And I can add initiatives manually. And we’ve got AI that will help you fill that out. But also I can use AI to brainstorm, right? I can say, what [00:40:00] roadmap initiatives might we be missing?

And the really key thing is that it’s keeping the human in the loop, right? It’s not replacing you. It’s not just putting stuff on your roadmap and then being like ship this it’s saying we’ve looked at your goals, what your customers were asking for, what’s on your roadmap, everything else. And based on that, here are some initiatives that might fit your roadmap and then put some in front of you so that you can say, ah, yeah, actually forgot about that.

Let’s include that and allow you to make that call. So right now it’s gathering all that information and it’s. And it’s going to generate some suggestions of roadmap initiatives here. So here you go. It’s come up with roadmap initiatives. And really importantly it’s already writing them in the question format.

Like, how might we do this? Or how could we do this? As opposed to trying to state features that you could go build. Because that comes later, obviously. And it’s, these are based on what it knows about our roadmap and what it thinks might be missing as well. So I can pick and choose which of [00:41:00] these I want to keep.

Let’s say we want to do that one. And maybe these two are both scope creeps. So I can just add a couple initiatives here. And now they’re added to my roadmap here. And I can start building them out in more detail. Start connecting them to the goals that are important. But also allows you to do more with it as well.

So let’s say I’ve got one of my initiatives here. There’s a problem that a lot of product managers run into, which is called falling in love with their idea. I know Teresa Torres talks about this quite a lot. When you’ve only got one idea on your road map and you’re saying this is the thing that’s going to solve this problem, AI can be used to help you think outside that box as well.

So it’s saying if this is the problem we’re trying to solve, we’ve only got one idea, what else might we do? And it will suggest, it’ll brainstorm ideas for you. Again, very importantly, keeping the human in the loop so that it’s not replacing you. It’s not replacing you as a decision maker. But it’s augmenting it and allowing you to update things based [00:42:00] on this feedback session that you’re doing.

Megan Saker: And we have a lot of customers who will just run this regardless, even if they’ve got an initiative which has two or three ideas, three or four ideas, they’ll just run this anyway, just as a, like a gap analysis. Are there any blind spots? Have we missed anything? 

Janna Bastow: Exactly that. Here’s some suggestions to come up with.

I can say, yeah, no, not really feeling that one. Yes, actually this one makes sense for us. Let’s add these two to our roadmap. So I can add these ideas or I can even use this refine button and it allows me to start using CoPilot as a chat thing to start talking to it and saying, hey, you know what, can you adjust this idea?

and actually tweak it to say something more like this and then add those ideas directly from there instead. So it’s using a combination of chat as well as instant prompting to do that. 

Megan Saker: Yep. Sorry, Jan, I was just thinking in terms of the background information [00:43:00] that the CoPilot is using to make these suggestions.

Is it worth just having a quick look at one of the canvases? 

Janna Bastow: Yeah, absolutely. Because 

Megan Saker: There’s also a button here across 

Janna Bastow: I gave us a peek at the canvas earlier. The canvas is like a one page or you capture your vision, your description, your value strategy. We have one at the portfolio level as well, which captures even higher level information, like your value proposition and revenue streams and stuff like that.

And so you can use AI here too. Analyze this and be like, Hey, taking a look at your vision. Is there anything that fits this vision statement very well? Is it too vague? Is it too specific? Is it not broad reaching enough? So we’ve taught it what a good vision looks like, and it gives good, helpful crit critical feedback, but also really helpful feedback in a format that allows you to work with it and adjust it.

So take a look at my vision statement. It says, yep, good. Thumbs up. It does go higher than that, right? So you can get an excellent result from this if you’ve actually worked on the improvements here. So it gives improvement suggestions. Again, I can use the refine button to talk [00:44:00] to CoPilot and ask it more questions about this or dive deeper.

Or I can go and adjust my vision and then run it again until I get it better and better. 

Megan Saker: Indeed tell CoPilot to do it, like Yeah, exactly. On, based on your suggestions, can you re, can you rewrite that for me? Exactly that? Yep. And the next tab along while we’re here, the OKRs oh, sorry. I’m sorry.

Yes, you can also help also get a CoPilot to, to help you here, right? Yes. If you’ve got a broad objective. You can help with, CoPilot can help with key results generation. 

Janna Bastow: So who here is in the middle of or has just finished doing key results for the year? Hopefully we’re done setting them for the year as we’re into February 3rd, but I know some people are still doing it as of last week.

You can add product objectives and you can use the key results brainstorm tool. to help you figure out key results. And it’s basically saying, if I want to generate some key results, when are these ending? Let’s say these are the ones that I want to have done by the end of March.

And it’s going to, again, keep the human in the loop. It’s going to check with us [00:45:00] first, but it’s going to suggest some key results. And what I love about this is that we’ve taught it to be, we’ve primed it to be capable of coming up with out, Come focused, leading metrics based on the context that you’ve set for your objectives, but also your wider product space.

And allows you to just quickly generate these. And if you don’t like them, you can throw them out and start again. But it’s basically saying, here’s some ways that you might want to measure this. And I can go, yeah, actually these ones aren’t relevant, but I want to add these two. There we go. And now it’s added these, which I can now use and adjust, and I can mark as whether they’re on track or behind, I can connect it to roadmap initiatives, or even generate initiatives from this.

So it allows you to go back in the back in the loop and really could just quickly come up with ways to solve the problems that you’re outlining. Yeah, 

Megan Saker: I’m just conscious of time actually and thinking about what, so someone’s just asked about user stories, and yes you can, but there’s a button.

There’s a button for user 

Janna Bastow: stories, who wants to do user stories? There we go. You can take this and it will [00:46:00] create user stories, so again, I can add them manually or I can generate them, and I can even choose. What format I want them to be generated in you can see here, the, I was using the, as a, I want, so that format.

And so it created those. These are all users, these are all AI generated that I’ve added previously which then connects to my idea. I can drag and drop them to reorder them. I can add them to the roadmap, push them onto development, or, of course, edit them and make sure that I’ve got something that’s perfectly target for our own product.

Another thing you can do here is I can take an idea and I can say, Hey, CoPilot, can you either generate the idea description, right? I can either flesh it out from the stub that I’ve written, the bullet points, or I can see whether this idea aligns with my product. So I’m just going to see if it aligns with my CloudWave Hub product, and again, it’s like a coach.

It’s going to help me identify whether this idea is any good or not and give me feedback on whether it does match. So really helpful to think of this as not just something to generate work, but to check your work, to [00:47:00] sense check to help you and your team make sense of whether these ideas work.

So it’s being honest with me, it’s saying this sort of aligns, but actually here’s some things you could do to make it clearer, or maybe you need to rewrite this idea, or maybe it needs to sit out compared to other stuff. And again, I can refine it, I can ask for more information from there.

But we also wanted to look at asking CoPilot stuff straight up, right? Yeah, we do. 

Megan Saker: There have been a couple of questions in the chat about feedback. So again, you will find buttons that’s CoPilot on any feedback entry for CoPilot to summarize feedback just straight off the bat. And we’ve got our signals tool that will analyze the entirety of your feedback and give you the themes at any one time.

Janna Bastow: Yeah, that’s really key. So it can take this big pile that I’ve got here. This is a big transcript. It might be a big long thing. I can do a back and forth email summary and it can basically just say, you know what, here’s what this person’s talking about. Let’s, some good, some bad.

Let’s add this to the summary. And now I can take [00:48:00] this and I can, select this area and connect it to one idea or select another area and connect it to another idea. Or I can allow AI to help me make those connections as well. And I can do that at the individual feedback level or at the entire feedback level.

I can say I’ve got a hundred pieces of feedback. What’s happening? BAM. And it turns out a lot of people are complaining about the dashboard usability and other people are talking about file formats. So it’s helping to dig stuff up and find stuff for us. But then I can do stuff and dive even deeper. Megan, we had a couple prompts there.

What did we want to ask it, 

Megan Saker: Yeah, we do. So we’ve talked a bit about how you can help so how Co pilot can help you with best practice stuff, can field your stakeholder questions. Let’s have a look. What about a stakeholder question? Yeah. Imagine you, you are a stakeholder these days, Janna, right?

You’re the CEO. 

Janna Bastow: Exactly, so let’s say I’m asking, So what are we actually going to get done this quarter, right? Give me the details. I’ve just typed this in, and I’m asking Co Pilot to answer this for [00:49:00] me. And this can be asked by you as a product manager on behalf of your team, or, just a prod pad in general.

Or it could be something that somebody in your team logs in and asks. So all those things that you normally beg Tapped on the shoulder for and all these questions, and you’ve got to go stop what you’re doing. Find the info. It’s actually already here. It’s now taking a look at what’s on our roadmap.

That’s going to be completed or due for completion this quarter. And it’s digging up those details. The direction of those initiatives and pulling that insight out for me, but also allows me to ask follow ups as well. So it’s going to go out and find that information. There we go.

And it’s now saying all my road map, these are the things that are happening and here’s where they sit. And now I can say, cool, can you tell me more about that Facilitate Team Innovation? It links to it for me, or I can ask follow up questions. Can you tell me which ideas are on there?

Or how things are going with it? And it’ll dig that up. 

Megan Saker: The other thing, of course, is with ProdPad, you get a free and unlimited reviewer. Users so you can invite anyone in your organization without having [00:50:00] to pay for a seat to come into your ProdPad and be able to use Copilot for their questions so they don’t come to you as much.

Yep. I’m asking 

Janna Bastow: Now, basically what it’s doing is connecting the dots for you, like as a product manager, you always have all this stuff in your head or in a bunch of sheets or tools or something like that. You’ve already got it all in ProdPad. So now it’s going to connect those dots for you.

It’s saying what are our customers asking for that isn’t on a road map? Or maybe something like, Hey, could you summarize everything that customer X over here asked for, or customers who look like this asked for, and then tell me, what sort of things we should consider on the roadmap, or what I should talk to them about next time I jump on a call with them.

So this is useful for anybody interacting with customers: your customer success team, your sales team, your product people, of course making sure that everyone has this insight, but is also able to ask the wider questions.

And there we go. So it pointed out that [00:51:00] does include some stuff, but some gaps here. We don’t have anything around file formats. And so customers are complaining about that. Here’s stuff that we want to fill out more from there. So it’s allowing me to see this information, but then also dig in and ask follow up questions to this.

So I could say, Oh, could you tell me more about those file formats or who’s asked for it. And it’ll dig that information up. 

Megan Saker: Yeah, and of course you can ask ProPilot for anything best practice wise that you want to know, if you’ve come across, if you’re about to start a new sort of product management task, you haven’t come across it before, you want to sense check your approach, or maybe you want to like onboard a new teammate and you want to explain, I don’t know, how to write user stories.

You can just, CoPilot can just tell them, rather than you having to outline it all. 

Janna Bastow: Exactly that. I recognize that we’re just about at time, but I want to leave time for some questions. We might go a little bit over today, because there was a lot of interest in this. Megan, do you want to help me out by picking out some of those [00:52:00] questions?

Megan Saker: Yeah, do you want to jump back over to the deck, Janna, because I just want to point out, I think the best way to fully understand what CoPilot can do is for you to get in and try it yourself and to have a play. So if you start a free trial of ProdPad you can just go ahead and do that.

There’s no credit card or anything. Create a product and then throw in whatever words you have about your product into that canvas. So you could put in a value a value proposition statement or a vision statement or whatever. And you could even, oh, cause this is something that ProdPad that CoPilot can do.

It can import your existing roadmap for you. So let’s say you’ve got a roadmap in, God forbid, a slide deck or a spreadsheet or something. Give it to CoPilot. Or just take a screenshot of it. Just take a screenshot 

Janna Bastow: of your old roadmap and then help you turn it into a ProdPad Now-Next-Later roadmap within just the chat function.

Megan Saker: [00:53:00] Yeah. Just give it the image file and say, can you add that to the roadmap for that product? So do those things. And then you’ve given a bit of. CoPilot bit of context and then just, and just have a play just set it to work. We’d love to know what you think of it, obviously. And yeah, that’s, we’ll recommend you all go and do that.

But if people are happy to hang on for another five minutes, we have got a whole bunch of questions here to ask. Okay. Janna, I have, Simon has jumped in and answered a bunch of them for us. Yeah, 

Janna Bastow: There was a top one there that multiple people have asked. Ian asked that question, including commercially sensitive information.

So happy to address that one. So the really key thing is that when you’re using ProdPad’s CoPilot, this information is not going back to be used for training information, right? Think of it as this closed environment that you and your team can use. to work on your own product, but none of it is used within the training [00:54:00] data or other accounts or anything like that.

It’s all specifically cut out for yours, which is different than what you get with ChatGPT, right? ChatGPT, by default, that information is used for training data which obviously we don’t want. You don’t want to put your whole strategy and all that other information in there because that could then pop up in somebody else’s answer.

Yeah, exactly. 

Megan Saker: But neither does CoPilot not store things. It does build up a memory, but it’s all stored on our side. It’s not given back. It’s for your own accounting, which is really 

Janna Bastow: key. Exactly that. Exactly that. All right. Any other questions that you want to jump into, Megan? 

Megan Saker: Let’s have a look.

There are a lot. I think what we’ll have to do because we are out of time, we’ll do a couple now, but I’ll curate these questions and may include some of the more popular ones and the answers in the email that we send round. Steve’s got an interesting one.

Where is AI and product management going to lead product managers to in the future? So whilst it improves [00:55:00] productivity, it seems to me that the role of product manager has to change. What do you think this would look like? 

Janna Bastow: Oh, love this question. Where I think product management is going is certainly not going to be replaced by AI.

But what it’s doing, it’s taking a lot of the work that was hard for us. Today and years ago and making it super simple, right? Creating specs or creating the product itself used to take tons and tons of hours of work, right? Just thinking about how long it took to spec something out for a login page or for a tag system or something like that.

It would take pages. Pages of details, and you’d have to be really specific about it. Hand that over to development and work to get the right thing out. Whereas, in the future, product managers should be able to simply define what it is that they’d like the change to be, and AI can make that change for them.

And it makes it so that product managers are very much less technical and much more user focused. They’re much more problem focused. They’re able to identify what problems the users need to solve and create the different variations that can be tested [00:56:00] and put out to life with very little effort.

It changes the role from a creator role into much more of a curator role. Where you’re basically saying, Hey, I’m talking to all these users. I’m understanding these are the big problems that we could solve that would be convincing for our teams. My role isn’t to chase some specs out the door and get this thing launched.

It’s to decide if all the different variations we could build, which one is the right one that’s for our business. And so it will take somebody to understand the market and to make those decisions. I’ve challenged the team already. We have a button within ProdPad that says take this idea.

The idea and the specs and push it to development. And it creates this really neat two way integration with your development tool. So if it pushes to Jira, it sends it to Jira. Now the developers can work on it. But put it this way, how long until we can change that to a push to production? Right now, there’s probably a lot of stages that are going to happen before we can push an idea to production code.

But in between that, it’s things like what can we do to help create a [00:57:00] prototype out of this? We’re talking about AI prototyping. I don’t think that’s too far on the horizon where you can say this is a cool idea that I’ve worked with AI to help create and I validated with users. Now help me create versions that I can test live, right?

Just make the prototype and then I’ll decide. And then from the prototype, you’ve got way more insight, way more information that developers can code the final production code on. But in the future, we’re going to be able to just push our ideas to production. And that’s going to be mind blowing. I can’t wait to see what we can do with that.

Yeah, absolutely. 

Megan Saker: What do you think, Jan, more questions or let’s take one more 

Janna Bastow: And then we will make it a wrap. 

Megan Saker: Okay. All right. Great. So Marvin asked can you create a user story using CoPilot that you can then add to the backlog? And can we show an example? So hopefully, obviously we showed how we can use CoPilot to generate a user story.

In terms of pushing it to, to, to the backlog what you can do with CoPilot is generate ideas. And then the [00:58:00] user stories to go along with them. And then prod pads workflow capabilities and integrations with your development tools means that then you can push those ideas. And through when they’re ready, when you’ve validated them, they’ve worked for your process and they’re ready to build.

Janna Bastow: So in short, Marvin the future is here. You indeed can do that, create an idea, break it down into user stories, and then press a button and it syncs to your development backlog from your product backlog. That might be all the time that we have for questions today. Cause I don’t want to leave people out too much, but we’ve got lots of questions that are left over.

So we’re going to try to tackle these as a follow up. segments, a written segment tackling these because these are really key things that we want people to know and understand. And we can go from there. 

Megan Saker: Yeah, absolutely. So thank you everyone. Thanks for coming along. Do look out for the email in your inbox.

Give us a day or two. We’ll get the recording there and some of the extra [00:59:00] resources. In the meantime, hop into ProdPad, start a trial, have a play and absolutely let us know what you think of ProdPad. 

Janna Bastow: Amazing. I’ve absolutely loved hearing from everybody today and look forward to chatting to you more about this in the future.

Until then, bye for now. Thanks everyone.