 I was recently asked to come up with a Sprint plan to identify opportunities within the climate tech space and I thought I might try to throw it at chat dpt and see what it came up with. So I took a previous design Sprint that we ran and identified some of the key components and asked GPT using this design Sprint as an outline to come up with one related to climate for me. So this was the initial prompt that I gave it, I'm planning a two week long research Sprint for Mozilla. My goal is to identify one to two business opportunities in the sustainable tech and climate tech business verticals. I need you to generate a Sprint plan that includes the following. Sprint goal, a list of the most profitable market segments, a list of research on the top five ways companies can make money through sustainability initiatives, the most likely target audience, a list of three how-might restatements, how to do product concepts based on those how-might restatements, a user journey that walks through the product concept, a testing protocol and a list of competitor businesses, verticals and products. It very quickly returned all of the things that I asked for in a single response and what I realized besides the fact that it can't number very well was that I was thinking about the problem very broadly because I hadn't really specified much context. Who Mozilla is, what I'm trying to accomplish, I hadn't really narrowed things down. But the framework that it provided me was pretty helpful. I knew I was focused on the B2C model and wanted to give people tools to take their own actions to be more sustainably minded. So I asked it, what are five ways a software company that makes privacy and security products and family tech products can help customers improve their carbon footprint? And it gave me another list, use of energy efficient devices, tools for reducing energy consumption, encouraging the use of renewable energy, promoting sustainable practices and offsetting carbon emissions. So these to me are good sort of broad stroke approaches that Mozilla could be advocating for. Not necessarily a product, but they are some of the major kind of common buckets to which you might address climate change. So continuing to shoot from the hip, I said, what's a good target audience segment? And chat GPT told me, that's not a good question. I need to know more about the business model and the product and the market. But I gave it a stab anyway, environmentalists, health conscious consumers, cost conscious consumers, luxury consumers, et cetera. And then it actually made a persona for each of those, gave them a name, gave them an age, talked about their concerns. And so I chose the cost conscious consumer, Mary, a 25 year old student on a budget, but wants to make environmentally friendly choices. So now I told it to regenerate those how might we statements for Mary and provide an answer to them. So again, it came right back with an answer. But as you can see here, the answers were fairly generic and it still couldn't number. So I got more specific and this is where chat GPT has excelled, where other language models haven't been able to take into consideration prior context of previous questions, chat GPT can. So I said, considering the product lifecycle assessment, what are three tools or features we could give to customers to help them reduce their energy usage and browse the web more sustainably? It said, make an energy efficient browsing mode. It provided ways to do that. It said, allow it to choose the most efficient data centers. It said, provide a carbon offset program. So it really tried to provide some diverse and fairly creative options for this context I'd given it. So I said, cool. Take those three things and combine them into a product concept, write a user journey. I came back with this product concept. It gave it a name called eco browsing again, very generic, not very creative unless you ask it to be and it decided it would be a browser extension that helps internet users reduce their carbon footprint. Mary downloads eco browsing extension for a browser. She browsed the web. The extension automatically activates the energy efficient browsing mode, reducing the energy usage of her device. She visits website, the extension displays information about the energy efficiency of the data center. She can choose to offset the emissions and she uses this and over time she changes her behavior. So it's got a surprising understanding of both technical features and change in human behavior, which I thought was interesting. So I'm thinking, okay, we have a product. What are some competitors? And it went and found existing competitors to this product, green browser, carbon free, eco surf, carbon click and green glow. So I'm thinking, cool, it found some competitors. Let me go research those. And then I realized I hadn't switched its frame of reference to the real world and real competitors. As you can see here, search for green browser shows that it doesn't exist. Or at least I'm pretty sure this is not the competitor it had in mind. An old school browser for old school systems. Trying to bring back old internet explorer. Similar search in the Chrome web store for, what was it, carbon free. Turns up nothing. Eco surf. Also, not a thing. But points for creativity. But I didn't at that point check those references. So I said, are they profitable? I said, I don't know. I don't have access to the financial data. So at least this time it admitted what it doesn't know, but that is a known problem of chat GPT that it will invent answers. When it doesn't know their veracity. So note to self, always follow up. I switched gears. Said, all right, back to those 10 questions. What can I ask users about eco browsing concept to figure out? Are they willing to pay for this thing? What are their concerns? It's a generated list of questions. This time a little more relevant and useful. And these are foreseeably something I could consider sticking into a user testing session. Now I switched more to sort of business strategy. I said, give me a business model can. And at this point, I was pulling from books I've read on lean UX and lean startup. So I didn't even have much context for what a business model canvas includes. But chat GPT didn't really need that. I said, okay, here you go. Keep partners, activities, resources, value proposition, customer segments, channels, customer relationships, revenue streams. And with respect to revenue streams, it actually said, hey, this can generate revenue both through subscriptions for the extension and fees from the carbon offset program. It gave me a SWOT analysis. It pointed out, hey, this has those two revenue streams. That's not always common. It recognized the societal benefit of carbon reduction. It said, hey, some people don't necessarily care about reducing their carbon footprint. It said, hey, you're going to potentially have a dependence on browser providers. Luckily, I know of one. But it said, hey, this is a growing market, and you can leverage partnerships with other companies in the sustainability space. You can expand into new markets, such as mobile apps and other software. It acknowledged there's other competitors. It said, consumer preferences might shift and technologies might emerge that will make this easier, that would make your concept obsolete. So not necessarily novel revelations, but good things to keep in mind as a sort of checklist as I work through this sprint planning. I said, okay, this time I don't even know the name of the framework, but what is a framework I could use to evaluate product concepts? It said, oh, here's a thing called triple bottom line, never heard of it. To use triple bottom line, you can ask the following questions. What are the financial costs? Will it generate revenue, reduce costs, have other economic impacts? What are the social aspects? What are the environmental aspects? By considering these three dimensions, you can evaluate the sustainability of different product concepts and determines which ones align with your values and goals. Cool. So now I'm thinking, why stop here? Let's let this thing keep doing my homework for me. Generate a lean UX feature hypothesis template. I don't even know if I'll use it. Here it is. Featured, problem, assumptions, hypothesis, metrics, success criteria, experiment. Again, all things I can kind of outline in the sprint plan to potentially implement to identify these product goals. I'll say, generate a lean UX hypothesis table for the concept that I mentioned to you. So again, giving a previous context applied to current responses and it generated this table and I even put it in table format. And it outlined each of the features that it had mentioned and said, here's the problem, here's the assumptions, the hypothesis, metrics, success criteria, and experiment. So this is simple. I haven't really looked at the success criteria like at least 20% of users visit websites hosted by energy efficient data centers. More frequently, that sounds somewhat reasonable. Carbon offset program. At least 10% of users participate in the carbon offset program. I thought this was super cool. But can you generate a matrix? Apparently it can. Here it is. It generated a prioritization matrix for each of the three features, energy efficient browsing mode, risk. Users do not activate the feature. Feature causes performance issues. Users do not notice the difference in energy usage. And it took a stab at the probability, impact, and priority of each of the three features. And it took a stab at the probability, impact, and priority of those things, which I could review and change accordingly. Same thing for data center. Same thing for carbon offset program. And then I'm thinking, well, can it do my marketing for it? Generate three to five calls to action to entice users to the extension. Join the fight against climate change. Download eco browsing today. These are all very aspirational and not necessarily language that I would try to approach my target audience with. So I said, hey, give me an alternative. Include something that allows them to learn about their carbon footprint, why it's important, how eco browsing can help. And it actually improved in my opinion. It said, find out how your online activities impact the planet, discover your carbon footprint, and learn how eco browsing can help you reduce it. So much more approachable, much more human, and it just required some additional context. I said, well, not quite right, but let's throw in an actual fact about the impact of an average user's browsing behavior. I said, did you know the average internet user's browsing activities contribute to approximately one ton of CO2 emissions per year? Find out how you can make a difference. And I was like, well, I don't know. Is this one of those things you made up? Sight the source. And it did. It said, here's a 2019 study published in the journal Environmental Research Letters, and it gave the citation. It gave a link. I went to that link and it was actually a dead link. So I said, oh, this is a 404. I need a link that returns a 200 response. And it said, sorry about that. And it modified the link. And it also said, alternatively, you could just search for carbon footprint internet use. So I said, OK, moving on, who could we partner with? And it said, renewable energy projects, browser providers, sustainability focused companies, data centers, environmental organizations. It gave a few examples of those. And obviously, it didn't get the hint that we are Mozilla. I'd have to figure out a better way to make that clear to chat TPD because it suggested Google, Microsoft, or Apple, which we, of course, wouldn't probably partner with. It said, for those partners, identify one company that we best to partner with. And I gave it some criteria. I said, it should have conducted a greenhouse gas inventory. It should have the lowest GHG emissions in its class, a net zero target, or at least the best track record for sustainability in its vertical. I don't think it really understood that last one because it didn't include any of that in its response. But it did identify, this time, real competitors or partners. It said, here's five companies, Carbon Conscious. It gave a website. It said, yep, I checked. It has a greenhouse gas inventory and a net zero target. And this time, the browser provider was Mozilla Firefox. Yay. So we've got Mozilla. We've got Carbon Conscious, Patagonia, Google as our data center, and Greenpeace as our environmental organization. So a real hardliner. And I said, OK, how about a Sprint retrospective? It gave me a Sprint retrospective template. What went well? What could have gone better? Action items. I'm like, OK, this is starting to look like a real comprehensive Sprint plan that I could modify and hand over to my boss and see what he thinks and give it to my team for feedback. So obviously, there's a lot more to the work than this, but it did get me going really quickly and kept me from reinventing the wheel on a few things, as well as provided a few creative solutions. I continued in this vein a little bit to see what chat GPT could come up with as far as research institutions doing innovative things. I wanted it to try and think outside the box a little bit. It pointed me to the Climate Browser, the GreenWeb Foundation, who we're actually partnering with, Carbon Conscious, and the Center for Sustainable Energy, UC San Diego. I don't know why I chose that one, but I definitely would be interested in following up on what they're doing. Optimization of web content delivery. Let's find out if that's true. I said, how about a project roadmap? It gave me that. I said, what are some analogs in other industries? It gave me that. I said, how can we reach customers to kind of promote this thing? And this was somewhat generic, but social media, online advertising, influencer marketing. I said, write me some AdWords copy. And it gave me some more fairly aspirational, very enthusiastic calls to action. I said, how can we differentiate? And I gave me an answer for that. I said, I don't really, these are a bit generic. Pretend you're a graduate level machine learning researcher. Now give me some innovation ideas. Think outside the box. Generate ideas that few people have tried. And this was actually really interesting, predictive content delivery. Figure out what the user is going to be looking for before they look for it. And depending on what they need, you can preload it, reduce the energy required to load the pages. Energy efficient routing, edge computing, renewable energy powered data centers, and adaptive streaming. So I could keep going down this line of questioning. I think this is a pretty cool tool to sort of facilitate innovation. And if you try it yourself, let me know what you learn and if you develop any techniques for generating useful information. Thanks.