StreamSparkAI

speaker 1: I have to read a lot of research papers for my job. Sometimes I get several sent to me in a single day. And even though I love hearing about everything new happening in my field, there are times when I can't get through every line of every paper. When that happens, I'll upload one of these papers to a generative AI tool and start a conversation I'll prompt for the most important points in the research and then ask follow up questions on information that I'm curious about. It's helped me get familiar with a lot of topics quickly because it's like having a creative on demand assistant that I can have a conversation with. In this module, you're going to learn about the best ways to prompt gen AI tools to help you summarize text, write better emails, level up your brainstorming, build schedules effortlessly, and take meeting notes instantly. Now let's learn more about what you can do with those prompts and how they can benefit you at work to boost your productivity and Spark your creativity.
speaker 2: Hi, I'm Clara, and I'm a product marketing manager on the Gemini team, and I' M1 of your instructors. Before I joined Google, it was a startup founder. And as any founder can tell you, one of the most important and challenging parts of running a startup is getting in touch with investors. I sent out a lot of emails, and sometimes I wouldn't get a response. I spent a lot of time coming up with creative follow up emails in order to get their attention. Generative AI is really good at that sort of thing. GenAI tools can help you draft emails, adjust your tone or style of writing, and give you ideas on how to change your approach. Here's how we can put that to work for you. Let's try this example. Using our prompt framework, we'll add the task, provide context, then evaluate and iterate. This is a useful exercise whether you're a copywriter working on an ad campaign, a speech writer drafting talking points for an executive, or gym manager communicating a schedule change. Let's start with the persona, I'm a gym manager. Next, we'll add in the context and we have a new gym schedule and specify the task, write an email informing our staff of the new schedule. Highlight the fact that the Monday, Wednesday, Friday cardio blast class change from 7:00a.m.to 6:00a.m.. And let's add even more specificity, especially about the content and format of the email, which could include language or tone. Make the email professional and friendly and short so that the readers can skim it quickly. And now let's just paste in the new gym schedule. Here's the new schedule. Okay, this is a really great email to send to the staff. All the information is there about the schedule changes, and it's short and easy to read. But did you know you can ask Gemini to adjust this same output for a new audience? This would be useful if, for example, you wanted your output to be more casual for a small, intimate team, more formal for the executives of the team, or even for a completely different audience. Let's try it. We're going to iterate on our prompt asking Gemini to write an email to the gym members in a more casual tone. We're asking for a new version here with new specifics and context. Now write a concise, consumer facing email with the same information for our gym members and clients. Let's get even more specific. Explain that this new schedule will help us better serve them. And now let's make sure we're clarifying the new tones, make the tone fun, friendly and motivating to reflect the gym's energy and enthusiasm for our members health goals. Include a pun about lifting weights. That's more like it. What a fun email. And it still has all the same information about the schedule changes. With just one extra prompt, we completely altered this email for a new audience without losing the important information. Can you think of a time when you were having trouble figuring out how to get the tone right in something you've written? With some iterations to your task and context, you can tweak your writing to land your message how you want it. Whenever you're drafting any sort of communications with the help of a gen AI tool, make sure the outputs are factual and adjust your prompts so that you receive the information that's most helpful to you. Whenever we hire someone new to join my team, we have a meeting to welcome them. What I've started doing is prompting generative AI for suggestions on fun icebreakers that are engaging even over video cop. I can even generate personalized icebreakers by including fun facts about my new teammate in the prompt. Whether you're leading a team, bonding experience, or thinking of ideas for a work project, you can use gen AI to generate ideas and brainstorm solutions to help you make more informed decisions. Remember, thoughtfully create really excellent inputs. You need to add your tacontext and include relevant references. Then you need to evaluate and iterate. In this lesson, you'll learn how to write prompts to generate ideas with Gemini. Consider this. You need to plan a launch for a new video game. Gemini can help you with everything from brainstorming sales strategies to managing the launch timeline. We'll start with the persona and include the context as well. I'm a marketer for a well known video game producer known for creating immersive story based online video games. I'm planning the launch of a new medieval fantasy role playing game that follows the path of a Young protagonist searching for their missing partner. The game's primary audience is Young adults. The game is reaching the end stages of development, and I need help creating a timeline before it goes live. Then we'll specify the tasks and include the formats, provide a rough timeline for the year leading up to its launch, also provide some pre launch ideas to help generate buzz around the game. Before we check the output, I want to highlight a few important things about this prompt. We've set up the persona. Your company launches video games in the context of medieval fantasy role playing game launch. We've also specified the task, create a timeline for the launch, and provide pre launch ideas to generate buzz. With all that in mind, let's review the output. This is a great start. Sign me up for the contest. While these are cool pre launch ideas, they could use in a little more detail, maybe your team is hoping to promote this game around the world. Let's give Gemini more context and ask us to update the launch events with a global audience in mind. Since we're already in conversation with Gemini, we'll follow up on the prompt we just wrote by adding now provide re launch ideas that will appeal to a global audience of Young adults. There we go. These ideas are suited for a global audience and not to mention seem really fun. I love the idea to include cultural references in the game to relate to players from around the world. Evaluating and iterating is like having a conversation with the gen AI I tool, continually guiding it to generate our desired output. We can all use a little help generating new ideas. How can gen AI I help you with brainstorming at work? Okay, spreadsheets aren't my favorite thing in the world. I'm a creative marketer, so I tried to avoid cells and formulas as much as possible. But there are times when I really need to use spreadsheets to create tables or track a project timeline. So I've started to use generative AI to help me, and it's made creating those tables much easier and less time consuming. No matter where you work, being able to easily turn information into an organized plan can be a massive help. Think of a schedule or project tracker. These documents are so important because they keep team members aware of their rules, les and responsibilities, and keep timelines and deliverables on track. But building them can be a chore. Let's explore how gen AI can help you speed that process along. Let's try this example. You need to create a schedule to plan your employee staffing each week, and want to prompt Gemini to create a table. Here's our task format and context. I have ten employees. Their employee numbers are one through ten. Create a table that tracks weekly staffing. Create columns for day name and shift, morning or afternoon. We also added constraints. There should always be two employees scheduled for the morning shift and two employees scheduled for the afternoon shift. Employees should not be scheduled for a morning shift on the following day after they were scheduled for an afternoon shift. Employees should not be scheduled for both the morning and afternoon shifts on the same day. Every employee should have roughly the same number of total shifts per week. In just a few seconds, gem and I created a staffing schedule in a format that's easy to track. The output includes the constraints we added when prompting, like making sure no one has to work a double shift. Even better, we can export the schedule to Google Sheets to continue working on it. After logging in, we can prompt Gemini in sheets to summarize our schedule, create a new schedule, or even create formulas to quickly analyze the information. But remember to avoid entering any sensitive information that you wouldn't want a tool reviewer to have access to. In this case, we wouldn't want to enter any specific information about employees. That's why we use anonymous employee numbers. Always consult your organization's policies regarding sensitive data and gen AI use before leveraging a gen AI tool for work. My team and I work in Google Docs on a daily basis. Sometimes when I'm editing, I'll realize the writing could be improved, but I'm not sure how to approach it. Instead of staring blankly at my screen, I can prompt Gemini and docs to pull out what I need to suggest something new. To do that, start by logging into docs. You'll need a Google account for this, let's try an example. You receive customer feedback that your company's return policy is too long and confusing. You can paste the return policy into a doc, then prompt Gemini. From there, you can enter the following prompt. Let's specify the task and context, summarize this content to write a clear and concise product return policy, and outline five steps for customers to take in sequential order. Now you have a solid return policy to reference when assisting customers as well as a list of ways customers can begin their returns. There's even more you can do to save time and boost productivity with Gemini and docs. Some days are just so packed with meetings and people and action items in those meetings that it can be hard to keep track of who said what or what you're responsible for. Think about a long meeting you had recently and just how much there was to cover. You might want a transcript, right? You can create one writing Google meet. From there, you can turn to Gemini and docs to help you keep things organized by creating meeting minutes, outlines, a bulleted list, and more. To illustrate how it works, we'll use a transcript generated by gen, AI, I and open up docs. Now let's prompt Gemini making sure to ask for exactly what you need. Using our prompt framework, thoughtfully create really excellent inputs. First, we'll add the task, create meeting minutes for this meeting, then specify the format section the meeting minutes by speaker, and create a bulleted summary of key points made by that speaker. If applicable, highlight action items for each speaker. The meeting notes might be more useful if formatted in a clearer way with more defined roles and responsibilities from the meeting. Let's iterate on our prompt. To achieve that, we'll use the same task.
speaker 1: create meeting .
speaker 2: minutes for this meeting, but we'll change the format and get more specific with what exactly it is we want from the output, create sections based on key points extracted from the minutes. Attribute each key point to a speaker, and then assign action items to the speaker who is most appropriate to complete them. Much better, the action items are clearly displayed, and you have the most important takeaways from each attendee. Now you can use these notes to align with your colleagues on a project, share updates to keep stakeholders informed, help your teammates complete the responsibilities, and monitor the progress of deliverables. When you're doing research, writing a report, or investigating market trends, you're bound to come across some especially complex or lengthy materials. Luckily, a generative AI tool can summarize long blocks of text, extract key insights, and even transform them into new resources. You can also specify your preferred format in your prompt, so you get exactly the type of output you want. What's the report you've been meaning to read, but you haven't had the time to you want to summarize it and pull out key insights. But since the report is so long, you'll have to use a gen AI tool with a long context window. A long context window allows for the processing of a large amount of information in different formats at once. The long context window also enhances the gen AI model's ability to understand and generate more coherent and contextually relevant responses to our props. We'll test this in Google AI studio. Let's open up AI studio and sign in with a Google account. Notice how we can give our prompt a name. This can make it easier to find our prompts later, and it can be useful when sharing prompts with others. Now let's start using the prompting framework. Here's the task, including the format. Summarize this report in an outline. Focus only on the most essential points. Pretty cool, huh? 44 pages of information distilled down into a few points. Let's call this prompt report summary. Long context windows aren't available on every gen AI I tool, so depending on the gen tool you use, it may not be able to analyze long documents. Gen AI technology is dynamic and ever changing, so always double check what features and capabilities are available in the tools you're using. And here's a pro tip in the meantime, if you're using a gen AI I tool that can't handle a large piece of text and you need a summary, break the text up and paste chunks of it in at a time prompt to summarize each chunk, and then ask to summarize the summaries. Whether you're doing research or analyzing reports, getting a summary of what you're reading can help you understand the material in a shorter amount of time. Another pro tip for you, not only can gen AI tools summarize lengthy documents, but they can also explain them to you in different ways. By incorporating a persona, specifying who the audience is, you can personalize the summary for different people. For example, you can ask things like summarize this for someone who isn't great with numbers, or summarize this for a small business owner, or even summarize this using 80 ties movie quotes. Now that you know a few summarizing techniques, let's discuss some more advanced methods you can use to condense information with gen AI I tools. When you ask a gen AI tool to provide a summary, sometimes the initial output is way too broad. You need a bit more detail to make it truly useful to you. A technique called chain of density prompting can help with that. With this method, you get increasingly concise summaries and end up with a chain of refined outputs. Think of it this way. You're working on the perfect elevator pitch for your company. Your first drap is way too long and includes too many details, but as you refine it, it gets shorter and more direct, all without losing the important points you need to get across. Eventually, you land on the perfect elevator pitch quick, direct and effective. Iterating on our last prompt, we want to update the task and format, provide a shorter summary, focusing on the reports, top two to three trends. Mission accomplished. But maybe you want it even shorter. We can get more specific with our prompt to ensure the output not only provides a simpler summary, but is also really targeted to your needs. Let's go shorter. We'll ask for a single sentence. Condense the summary into a single sentence, an even more specific context, highlighting the most dominant trend. Nice. We kept asking it to refine further, and we were direct and specific about the desired output. We can repeat this process until we generate the most useful and helpful outputs for the task. We can even get ahead of these extra iteration steps and ask that the gen AI tool provides multiple, increasingly short iterations in the initial prompt. You might be wondering why we wouldn't get to the shortest output right away. Well, iteration helps us check for hallucinations as we go from something super detailed to an output that's more direct and to the point. It also helps us stay informed about what information is being taken out as the outputs get shorter. Chain of density prompting also works in the other direction. If the output is too short, we can also prompt to increase the length and add more detail. Remember when I ask you to think of a report you haven't found the time to read? Give these prompting techniques a try and discover how quickly you can generate a helpful summary.