You’re reading this because you know artificial intelligence (AI) has value, but you don’t know how to use it. This blog gives you the seven most common ways to use AI in everyday work.

These use cases can be used with any AI model including ChatGPT, Gemini, Claude, Grok, etc. Don’t know what those are? Check out our guide to the most popular AI models.

All of the use cases below assume you are familiar with prompting and the chat interface for AI tools. If you want a very basic explanation of the technology driving AI, visit our technical breakdown of the technology behind artificial intelligence and large language models.

Content creation

The highest return on investment for AI is content creation. Content creation doesn’t mean you are a “creative” such as a graphic designer, podcaster, or videographer. It means you create content of any kind. If you’ve written an email, you have created content. That’s almost everyone.

Content creation is easier with AI for one simple reason: it is more efficient to start with a rough draft than a blank page.

That’s true for anything. You could be creating an email, year-end report, text message, resume, product description, policy document, speech, sales talking points, or countless of other deliverables and it will always be easier when you start with something rather than nothing.

If this is your first time using AI, we recommend using the CASE format:

• Context: Tell the story of your ask. When someone asks you to do something, there’s usually a story that comes with it. What are you working on? What’s your role? Why are you doing this? Is there a deadline? You’ll get a better response if you include the context.
• Ask: State what you are asking AI to do. This is the most powerful part of your prompt. You can write a prompt the length of a book all about this deliverable you need to create, but if at the end you ask “give me a short compliment,” then you will get a short compliment! Ask and you will receive.
• Samples: Attach a document or point to a sample of what you want. This isn’t always necessary, but it can be very helpful if you have an established workflow.
• Examples: Tell the AI how it can succeed. This is not the same thing as Samples, but the two work together. A sample is just a document, but the examples give more direction. You can attach a document as a sample, but for examples you can say you like its writing style, formatting, or conclusion paragraph.

The context, request, and example don’t have to be a single sentence. You can prompt AI the way you would chat with another person — casually, with loose direction — so long as your overall message is clear. Here are some examples:

• I just started a new job and I want to introduce myself to my manager. I’m worried I’ll make a bad first impression. Write an email introducing myself and convey how I can provide value to the team. Make me sound professional but approachable.
• I am writing my goals for this upcoming year to share with my team. Last year we had layoffs and lost 30 percent of team of 26 people. I know there’s a big shift toward adopting new technology like artificial intelligence. Me and my team work in internal communications and some of my colleagues are thinking about how to implement AI into our internal comms process. Last year I was evaluated primarily on the number of communication deliverables I produced. I did roughly 36 communications deliverables last year. Considering our team reduction, I think it will be larger. These seem the most relevant concerns, but I don’t want to get tunnel vision on my concerns. I want to include some goals that are common across industries. Draft a goal statement in the format of bullets for each goal with a short description of the goal. Use the best practices and standards in the healthcare industry for this type of deliverable.
• My friend keeps inviting me to catch up. I don’t know how to tell them I don’t really value the friendship and wish he would stop inviting me. Write a text message response that is polite but indicates how I feel. I don’t want to get invited again.

You can also draft longform content, but remember some AI models will be resistant to output longform content. This is because AI companies want you to conserve tokens. If your prompt includes an outline or sufficiently specific instructions to give the model confidence in a longer output, then it will produce what you are looking for. Here are some examples:

• I am working on a financial yearend report. Create a draft of a typical financial yearend report using best practices and standards from other financial reports of Fortune 500 companies. This should include an introduction, a letter from the executive to investors, a table of contents, snapshot of financial numbers, revenue, expenses, then a breakdown of each specific department. We have 9 departments: soft drinks, carbonated water, bottled water, confectioneries, salty snacks, sports drinks, tea, coffee, and energy drinks.
• I am preparing for my Dungeons & Dragons campaign. This next section is going to be a Wizard’s Tower with 4 different sections. The opening hallway, the combat arena, the upper archives, and finally a magical plane after the archives. Create a reference document for me to use while DMing this adventure. Include unique items and loot for each section. Include unique enemies for each section balanced for a level 5 party of 4 adventurers. Include description text for each room, item, and enemy. This should be a relatively easy encounter with a focus on world-building. You can take liberties with the specifics. This is a self-contained area.

Remember the output you get from AI is often a good first draft, but rarely a great final draft. This technology can help you do your job, but the work still needs your input to be high quality. It relies on your skills, expertise, and insight to take its outputs and make them good enough to use in your work.  

Summarization

The most common usage of AI is summarization. This is for one simple reason: no one wants to spend the time it takes to read something.

People don’t read because there’s too much to read. Research on how the average office worker spends their time suggests more than 12 hours a week is wasted on reading emails. That’s a third of your job. In one year, you’re spending 600 hours reading emails. This data doesn’t even include reading other documents. You probably read other things like status reports, draft presentations, memos, chat messages, text messages, etc. It’s not uncommon for “work” to be synonymous with “reading inefficient communications.” With all those hours going to reading things, when do you have time to do your job? For many people, that is their job. It’s not rewarding or efficient.

Take back your life by getting summaries from AI. Remember, the format of “context, request, and examples” are just as important when requesting summaries. If you’re attaching a 50-page document, the AI doesn’t know how to prioritize the information unless you tell it your priorities. Here are some examples:

• I’m reviewing a deliverable by my team. This is supposed to be a short blog entry that connects with the reader with direct language, practical examples, and plentiful screenshots. Provide a summary of the information in this deliverable and inform me if these requirements were met.
• Look at this email chain /[SUBJECT TITLE]. Give me a brief summary of events but include if 1) a request was made of me 2) a request was made of my manager 3) there is some sort of block to completing the project being discussed. I do not need details if this is a low-value back and forth with little information exchanged.
• I have been on vacation for 10 days. Review my inbox and provide a summary of the most important emails I received in that time. Important can be defined by: an email from my manager, a request made of me, a time-sensitive deadline of any kind, or any other definition of “important” you are familiar with. You do not need to inform me of other emails outside this criteria.

Two of the above examples are specifically for summarizing email, which is a unique strength for Microsoft Copilot. One of the strengths of Copilot is it can directly connect with your Microsoft 365 suite of applications — including Outlook. Check out our article comparing the differences of each AI model.

If you don’t have Microsoft Copilot, you can still take advantage of these prompts by attaching them to your prompts or simply copy/pasting them. This approach necessarily introduces “digital friction” into your workflow, which can be challenging depending on the model you’re using and the needs of your work.

Extraction

One of the quickest improvements to quality of life is relying on a technique called extraction. Extraction takes information from somewhere and places it somewhere else in a different format. It sounds simple, because it is! Extraction is most common when you’re taking information/data from multiple sources and compiling it for a new document.

For example, if you are assembling a report that relies on information from different places. This could be yearend financials for a company, analysis of housing costs in a market, or collecting data on consumer trends. The information driving these reports may come from 10 different sources, using 10 different formats, and may even use 10 slightly different metrics. One might report overall costs, while another reports costs per capita. One report might be a PDF and another is a spreadsheet. Collecting this information into one document can be incredibly time-consuming. Extractions saves countless hours of copy/pasting and reformatting.

• I am assembling a yearend financial report for my company. I’ve attached 4 documents detailing financials for 4 different departments: Sales, Operations, Customer Support, and Marketing. Extract the total costs from each document and create a table identifying the department, employee headcount, total costs, and total costs per employee.
•  

Classification

One of the greatest values of AI is getting neutral and unbiased analysis. You can use AI to classify the sentiment or tone of a document. This is especially valuable when you’re working with large datasets or when you’re looking for feedback on your own creative content.

Humans are naturally prone to a variety of biases and that can confuse your understanding of data. For example, if you’re looking at customer feedback on a new product launch you may see the first 10 reviews out of 100 are all explicitly negative. We know from established science that reading 10 reviews back-to-back is going to influence your interpretation of the data. You’re going to be biased to think the overall feedback negative, but that may not be true. The next 90 reviews could be positive, but those first ten are going to stay with you. Asking AI to analyze datasets allows you to overcome this bias and work for productively.

Classification is best used for datasets like the above example, but it can also be used for general feedback. An emerging use case for AI is mental health support to reveal insights about your own behavior. You can keep a personal journal for your own mental health, then provide this journal to AI and ask for feedback on reoccurring issues or challenges you need to address in your life. Of course, beware of the implications of providing AI personal details about your life.

Examples:

• I am reviewing consumer feedback on our new website. Review the attached document and classify the general sentiment of the feedback. Organize this output into positive, neutral, and negative. For each sentiment, include the top three topics/concerns brought up by these sentiments. Provide this in a bulleted list format.
• I’ve written a blog entry explaining 7 ways to use AI in everyday work. Review this blog and classify the tone and style of the writing. Provide your analysis of who this type of writing appeals to the most. Additionally, provide feedback on how the tone and style could change to fit the audience of small to medium sized business owners.

Translation

Translation is an underutilized feature of AI. Most people rely on Google Translate for their minimal translation needs, but existing translation services have their challenges. Any bilingual person can tell you how literal translations can fail to capture the true meaning behind slang or colloquial phrases.

Imagine you’re reading a letter from an international business associate who is happy to report their department “keeps money in the dark” because a literal translation of financials being “in the black.” Or when a word is used in one language that has no true translation in another language – words like [continue later]

The technology driving AI is called a large language model. We have a full technical breakdown of what that means, but the important thing to understand is it is based on language. This uniquely positions AI to translate to any language on the planet – well, most of them.

Use translation to expand the reach of your content.

• I’ve created a blog explaining the value of artificial intelligence, but I want to translate it to other audiences interested in this topic. Review the document I’ve attached and translate it into Spanish. Identify any sentences or sections that needed to be significantly re-written to retain their meaning after translation.

Editing

When it comes to editing with AI, remember to dream a little bigger. You don’t need an AI model to implement edits like grammar or spellcheck. That form of editing was already available in inexpensive products that don’t use AI technology. Editing with AI enables you to implement large sweeping creative changes. With a simple prompt, you can rewrite a deliverable to fit for a different audience.

Examples

• I’m rewriting product descriptions for my company’s offerings. These products were originally for young professionals and have a casual, hip, and affordable value. We’re pivoting to appeal to older professionals and we want to communicate seriousness, reliability, and high quality. Rewrite the descriptions of these products for this new audience. Keep the length of descriptions similar to the previous ones.
• My manager gave me feedback on this deliverable. Implement her edits. I’ve pasted her notes below.
  
  “Too wordy. Too many technical details. Use more storytelling at the beginning. Include a call to action between every third paragraph. This is decent but needs work!”

Problem Solving

The final use for AI is the most broad and has the highest potential value. Since this technology effectively unlocks all of the world’s knowledge, this means you have the opportunity to solve any problem in your work even work beyond your personal skillset and experience.

In pre-AI work, most workers specialized in a skill and relied on coordinating with others to effectively tackle large-scale problems. With AI, you can become a specialist in any discipline.