Back to Blog

How IWB Uses genAI (Part 2: Tips for Integrating into Your Workflow)

7/22/2025 12:00 am

In our first entry in this series, we provided a "cheat sheet" for which generative AI tools work best in different situations (and when you might want to think about switching to a new tool!)

 

In the second entry into this blog series, we present tips and tricks for using AI in your workflow, along with some quotes from our members about how they are using these tools. We acknowledge that this is not an uncomplicated topic! There are many downsides and nuances to integrating these new technologies – this will be the topic of part 3.

  1. Writing & Reading Emails
    • Consolidating Language: ever find yourself writing an email that you can tell has too much information and you’re overexplaining? Or you’re writing a blog and you really want to make sure people understand the time and place of when you would recommend using generative AI for email generation, but you want to make sure you get all the details right and include all the information you would ever need to provide them? Wait, should I put this paragraph through Copilot? Here are some ways that you can prompt generative AI tools to help you cut down on the text of your communications:
      • Give it explicit goals for the consolidated text (including: specific details/numbers that cannot be removed, goal length, intended audience)
      • Instruct it to maintain your current tone.
    • Generating content: What if you know what you want to say, but just can’t quite get the words on the page? Especially when reaching out to someone new or making an important professional connection where first impressions matter?
      • Give the model the roles of the audience and author
      • Include key points about what you want to say, as well as aspects you’d like to emphasize
      • Make sure you go through and fill in details wherever it might put generic terms, like [Your Name] or [Addressee’s Name] (or provide these in your initial prompt)
    • Changing the tone: Frustrated or anxious about a situation but want to ensure you’re projecting a calmer version of yourself in your email? Working on removing the “fluff” or undermining language from your communications (something that is often associated with women’s communication, but may not always work in our favor)? As with the previous suggestions, make sure you’re including:
      • Sender and audience positions
      • What content should be included
      • The desired tone of the writing (ie. make the tone of this email complimentary, but assertive)
    • Help deciphering the tone of emails: use AI tools to judge the tone of an email you received (may be particularly helpful as a support for neurodivergent users)
      Here are some examples of how specific prompting can help in email generation. Consider the following email "generated" by "Emily Smith," who is applying for a postdoctoral fellowship in "Dr. Johnson's" lab. This is her original email, but she knows it's not ready to send.

      There are a few typos and the tone doesn't feel quite right - to improve it, she might use the following prompt: "Can you make this email sound more professional?" and receive the following:


      Sounding a little bit better - but maybe a tad long? To shorten it, consider prompting "Can you ensure this email is professional and concise?" and receive:

      Or maybe the original, somewhat removed tone doesn't really feel as friendly as Emily likes to present herself as. To make the original email a bit more warm, consider prompting "Can you make this email professional but friendly?" and receive:
      Continue implementing different prompting techniques alongside your own editing work to iterate towards a final product that is striking the correct tone, but feels authentic to your voice!
  2. Refining Writing
    • Extracting/identifying key concepts from a large body of text: Generative AI tools can really shine when given specific (and known-to-be-factual) textual input. Some ways this might be useful are:
      • Asking the model to identify key concepts in your own work (this might also produce new wording that articulates your larger point in a way that you like)
      • Assessing alignment with mission or criteria – if you’re applying for a funding opportunity, you can use these tools to determine if your stated objectives or impacts are aligning with the posted description. Additionally, you could ask the model to pull out key concepts from the funding opportunity description to guide your application structure/emphasis.
    • Converting to reading age/level: While scientists are practiced at writing to other scientists, we can struggle when asked to write something in a different style, use a different method of communication, or tailor something for a different audience, such as a press release, consent form, or marketing tool. When prompting the model to update the reading level, include:
      • Target reading age
      • What content is necessary and what can be excluded
      • Preferred length
      • The context needed (word count, style, etc.)
      • Content of the work (as much of your manuscript as possible, or at least the abstract)
      • Request multiple options
      • [Note: while generative AI tools can do this automatically, if you would like to work on refining it on your own, but have some more difficult sections identified for you, Hemingway Editor is a great tool!]
  3. Generating titles, transitions, and topic sentences: If you’re struggling to come up with a title, topic sentence, conclusion, etc., prompt the model with
    • The context needed (word count, style, etc.)
    • Content of the work (as much of your manuscript as possible, or at least the abstract)
    • Request multiple options

      I use it for formatting. For example, I reformatted my CV and a SPSS variable code. I use it for generating creative titles for presentations. I also use it for non-work-related tasks like making dinner plans and grocery lists, which allows me to free up my mind for other things. I also will use it to gather non-work-related data. Like, I am going to Boston and want to know some fun things to do and have an itinerary. I think it helps minimize the time spent doing tedious tasks that are still important for me. I use it for these types of things. - Erin Lally

  4. Finding Information
    • In a text or file: You are busy! Sometimes you don’t have time to read a whole 70-page document to find the 1-paragraph section that you need, and maybe Ctrl+F isn’t quite working the way you need. Tools like NousWise (and others) will provide summaries or text identification from sources you specify. Use this by:
      • Uploading reference material
      • Asking for specific information from the uploaded reference material (some tools will limit this automatically, others will have to be prompted to do so)
    • New sources: Some new tools made specifically for academics can perform a highly sophisticated search for related sources. For example, in ResearchRabbit or ConnectedPapers, you can: 
      • Link a Zotero Folder / enter a paper title or DOI
      • Explore content linked by the AI tool (which will show how the sources are connected and grouped)
  5. Summarizing Information 
    • Synthesis: Want to get a quick summary/birds’ eye view of a new topic area?
      • Enter a research question (e.g., How does the gender of the researcher affect the field of biomechanics?)
      • Filter by publication year, impact, etc.
      • Tools like STORM will “show their work” so you can evaluate where the summary is coming from
    • Meeting transcripts: for generating quick minutes documentation or catching up on something you missed, using AI for meeting transcripts can allow for
      • A clear, concise summary of the discussion
      • Identification of clear action items
      • Highlighting key discussion points/next steps
      • [Note: We tested this in Zoom for this blog]
    • Interviews and surveys can often produce a large amount of dense information, and actionable insights can be difficult to extract. To use AI tools to help untangle these: 
      • Specify the topic
      • Indicate which key points you want summarized
      • Paste transcript/responses directly into the chat box, or upload a file containing the transcript/responses
  1. Understanding information: New AI tools, when used properly, may enhance the learning process. This can take the form of:
    • Summarizing information (and transforming it into written or audio tools)
    • Creating a personalized study companion/tutor – upload study materials, set learning level, and then “chat” with the “tutor”

      Generative AI speeds up my workflow by providing quick suggestions and breaking down complex tasks into more manageable steps. When I encounter a challenging problem, AI often helps me find a simpler approach. It has also made learning new programming languages much easier. For example, I learned R just by asking ChatGPT to guide me through different coding problems. - Asal Aflatounian

  2. Cleaning up data: Instead of using macros or creating brand new code, simply feed your data into the model and make requests about how you’d like the organization to be updated.

  3. Administrative tasks: Models like Goblin.Tools and other generative AI tools can give you a helping hand in organizing your day. Use them to:
    • Timeblock your day (just note what tasks you want to include, the timeline, and any meetings/breaks)
    • Create a to do list – Goblin.Tools can estimate the time for each task and break them into subtasks.

 

It is important to note that with any of these tasks, errors and hallucinations are possible, so be sure to check any responses and outputs from AI tools. Responses from AI are exactly that – there is no internal motivation or responsibility for the AI tool to be correct. Likewise, AI, while very advanced, can only respond to the prompt that it is given, the quality and specificity of the prompt will determine the quality and relevance of the response.

 

 

 

Authors

Julia Dunn, Ph.D. (she/her) (University of Denver)

Hannah Dimmick, Ph.D. (she/her) (University of Colorado Anschutz Medical Campus)