Education technology and game-based learning resources, tool tips, innovative pedagogical musings, and general "thinkalouds" of a Digital Learning Consultant. Follow my educational journey on Instagram, X , Threads and Bluesky @watsonedtech, my Facebook Page (facebook.com/edtechelixirs), and the blog hashtag #edtechelixirs. Note that opinions are mine alone, and reviews/mentions of edtech products do not qualify as an endorsement. Subscribe to the blog at bit.ly/signupedtechelixirs.
While Google's NotebookLM has taken much of the acclaim and headlines for how Gemini AI can transform learning, another free tool over the past year has been shaping up to be its contender: Google Gems. While Notebooks can offer audio and video overviews, infographics, slide decks, and seemingly new beta Studio tools every fortnight, Gems can offer a more nuanced kind of support. Gems are customizable AI chatbots you can load up with your own vetted documents, which you then can easily share.
For a short (12:08) overview on Gems, watch this video by Eric Curts (and more on Curts in a moment):
Recently, Google for Education provided a wonderful deep slide deck of resources titled "100+ ways to use Gemini in K-12 Education." One of the slides gives a short comparison of Gems versus NotebookLM:
A labeled screenshot of Google Classroom, labelled with the differences between Gems and NotebookLM. (Unless otherwise indicated, all screenshots are from this slide deck.)
As another way of comparing the two, here are some pertinent details:
Purposes. NotebookLM is usually for larger, deeper synthesis needs with multiple output opportunities, while Gems are more for specific, single outputs.
Amount of files. NotebookLM can have up to 50 sources uploaded; Gems can have up to 10. However, a possible workaround for Gems is below.
Kinds of files. Both can use many of the usual suspects: PDFs, Google Docs and Sheets, Word and Excel, images and videos. However, NotebookLM also allows website URLs, YouTube videos, and "copy and paste" sources, while Gems does not. Interestingly, however, Gems will always recognize and use the latest version of a Google Doc or Sheet attached; NotebookLM only knows the "instance" version of the Doc/Sheet when first uploaded.
Kinds of outputs. As mentioned before, NotebookLM has a wide (and growing) set of Studio tools, starting with Audio and Video Overviews. A Gem is more of a monotask outputter, only doing what it is set up to do. These do not have to simply be text-based lesson plans or an email, however! To take just two examples, a Gem could create coloring pages or a graphic novel.
Gems are great if you want to take the burden of prompt engineering off of the end user by writing a guiding "script," yet ensure a guard rail experience where the source of knowledge is based only on the documents of your choosing. In effect, you are being a light programmer creating a new app, but without worrying about having to write actual code.
Gems sit inside of Gemini itself, in its "Gem Manager" dashboard. You can see Gems you made and saved (which you can always edit and customize further), Gems shared with you (as you can receive or give sharing rights like any other Google Doc), and of course, create a new Gem.
Writing an excellent prompt is the heart of a Gem. For many an AI PD, I've shared Google's PARTS strategy on how to compose a more effective script (taken from this free course) by thinking of how to be more specific and detailed when thinking of each, um, part of PARTS. However, the original handout I've used considers PARTS from the first person perspective of the prompter ("I am a third grade teacher.."). For a Gem, it may be more helpful to consider a second person perspective, like the example below. In a meta moment, with a push of the "magic wand" button in Gems, you can also use "Gemini to rewrite [the] instructions" for you.
As I mentioned before, Gems can be shared with specific people, or made usable by all via its hyperlink. Indeed, Gems could theoretically be shared inside a Google Classroom with students. However, the use of Gems by students in your educational domain may or may not be allowed, depending on the settings determined by your Google admin. (For more, read this help article.) Dr. Jim Masters, superintendent of Henry County Public Schools (KY), launched a specific Gem math tutor "grounded in learning science" with students. Afterwards, over 250 of them answered a survey, and the results were noteworthy: "57% can now solve a similar problem independently" and "72% said [how the Gem demonstrated] breaking the problem into steps helped them most." While some students admitted they wished the Gem would have just given them the answer (therefore admitting the Gem kept productive struggle and didn't minimize their cognitive load), many also said they still appreciated, or even preferred, the way their teachers helped them work through a problem. The humans still matter in the loop. "Technology can scale feedback," said Dr. Masters, but "teachers develop thinkers. Student voice is now driving the next iteration — the same way a product team studies user experience."
When learning how to create a well-engineered Gem, a model can be helpful so you can peek "behind the curtain" and analyze its script. (They tend to be more complicated and "code-like" than a typical genAI prompt.) There are also moments when you want to save time and simply utilize a well designed Gem made by someone else. With that, we return to Eric Curts and his EduGems website:
The site is a collection of curated Gems, most made by Curts himself. The Gems run the gamet -- some educator-facing, some for students to use directly, and for various contents and needs. There's a category that really caught my attention (for obvious reasons) called Engagement and Games, with Gems such as "CYOA Story" and "Historical What If." Let's look at one called "Gamified Learning":
Along with a short description of each Gem, you have two choices. You can interact directly ("Use the Gem") or "Make Your Own Copy"; the link is a force copy that will automatically save it into your own Gem Manager. Note that once you have your copy, you can preview its prompt and potentially customize it.
Here is the Gamified Learning Gem "ready to use." Note that you have to give a Gem some kind of indication that you want to start -- typing "Hi" or "Begin" is fine.
When interacting with a copy of "Gamified Learning," you can see the Gem in editing mode. Note the full prompt on the left and the chance to preview the Gem on the right.
EduGems also offers tutorials, and a Google Form where you can submit your own Gem for potential inclusion on the site.
Google Gems offer a handy way of reusing an excellent prompt over and over and easily make a sharable chatbot based on key sources you trust. Even if you don't use Gems often, deconstructing a well-made one can teach you a lot about AI and prompt engineering. Thanks to Google for making another highly useful free tool. And special thanks to Eric Curts for curating an educational Infinity Gauntlet website of inspiring Gems!
Google has had a busy summer with plenty of updates, including a new product that I've finally had some time to play with: Google Vids. But before we get to Vids, let's briefly cover some other territory that educators will find important to know.
If you're a teacher who has logged into Google Classroom lately, you've probably noticed that your workspace now includes Gemini baked in, and more specifically, has multiple tools at your fingertips to help you more effectively complete tasks. It's clear that the popularity of the "custom bot dashboards" prevalent in several other AI learning platforms has encouraged Google Classroom to do the same.
[Update 10/14/25: Eric Curts created a deep dive video on the various pre-made AI tools inside Google Classroom, linked here (19:03)].
While the built-in capability of Gemini is currently only available for the teacher side of Google Classroom, Gemini itself is now available for students of all ages. Additionally, Google NotebookLM (perhaps the most pragmatic AI tool I currently use) will also be available for "all ages in the coming months" (per a Google Workplace Team email sent June 17, 2025). You can read more in Google's June 2025 blog announcement. Of course, your Google Admin may choose to keep this toggled off for certain age groups (or for all of your students!) and/or for one or both tools; it's worth asking what your district's plan might be. Remember that for educational accounts only, Google has made assurances that AI data and usage will remain private to the domain itself, usage will not train Google's AI machine, and inputs/outputs will not be human reviewed, as they first announced back in May 2024.
Speaking of NotebookLM, it's perhaps no surprise that it continues to upgrade. The output language can now be changed, and that can include the (in)famous "Audio Overviews." You've likely been wowed by the Audio Overviews feature (and if you're like me, also a bit disturbed by its capacity for human mimicry); customization can now allow you to make them shorter or longer in length, or give the Overview specific topics to focus upon. But perhaps the most dramatic leap for NotebookLM is the beginning of Video Overviews:
I was able to try this out on a personal free gmail account (upgraded features usually tend to chronologically lag behind for educational domains, so be patient if you don't see the feature yet). I picked a Notebook with two sources that were focused on the topic of flexible learning spaces: a Google Slide conference presentation I did back in 2021 about the then just opened K-8 school Marnel C. Moorman in Shelby County, and a 2018 article in Edutopia. Despite only having those relatively short sources, the generation for what turned out to be a seven minute Video Overview took at least ten minutes on a Friday night. While it's likely this will get better, or possibly be another incentive to go Pro (for faster generations?), I would caution you from expecting a Video Overview to load in the same short time it currently takes to generate an Audio Overview. Yes, yes, you might be asking, but what was the quality? Well...I was blown away, just like I was about Audio Overviews only a year ago. As I'm still impressed and mulling over the possibilities of what an Audio Overview can bring to the table for learners, the Video Overview comes along. It basically creates your very own short documentary of your uploaded sources. And like Audio Overviews, you can share or download the videos. (In order to watch the video via a shared hyperlink, you have to share viewing rights to the entire Notebook.) As always, the AI was not perfect, but I'll take some of the blame too. For example, my presentation had a quote from a different Edutopia article spoken by a teacher in a California school (attributed by the article name and hyperlink on the slide), but the video narrator mistakenly stated "as one of the MCM teachers Stacey Lamb said..," which, given the slide scanned superficially as is, seems a fair misunderstanding. Watching the whole seven minutes, I found very little else to nitpick, especially considering if my goal was to create a sophisticated summarization of my sources. Here's the Video Overview it generated for me, also downloaded and saved to my Google Drive:
Staying on the topic of videos: one of Google's tools getting some considerable press recently is Veo 3. This is another example of the continual evolution of text-to-video genAI in the marketplace; the outputs are longer, more sophisticated, and can now include AI-generated sound and dialogue. It's impressive, and brings more ethical concerns, but I mainly passed it by for practical reasons -- it's not cheap, and therefore not likely to be in a typical K-12 teacher's classroom anytime soon. (However, Google is offering a free year of their AI Pro plan to college students, if they redeem the offer by October 6.)
But if you are looking for a practical Google video tool for teachers and students, it's now finally time to discuss Google Vids. In a nice flip, this is currently available for all educational users at no charge, but is not available for free/personal Google accounts. Here's a quick elevator pitch to describe the tool: "An online video editor that's as easy to use as Google Slides."
Creating a new Google Vid is like creating most products in the Google Suite: it's an option for the New button in your Drive, or a choice on your waffle, or simply typing vids.new in your address bar. There are templates you can choose from if you want to create your project more quickly.
You can import media (such as Google Slide decks, audio files, and screencasts), make everything from scratch, or do a bit of both. A tip: I think it's easier to "compose" the skeleton of your video by starting in Google Slides, then importing it into Vids. However, note that anything animated, or clickable, will simply "flatten" once it gets into Vids (after all, you can't hyperlink inside a video), much in the same way that Slides downloaded as a PDF may lose the same features.
Once in editor mode, you can easily add transitions between "scenes," or lengthen/shorten a scene by dragging it by the end. You can also record a video or audio file to insert into your project, making choices like doing a voiceover a breeze. One of my favorite features of Vids is Script, located in the sidebar. You can type what you want to say and it will scroll like a teleprompter while you record your audio or video. If recording a video, this also means you can look directly into the webcam while you speak.
This screenshot of using a Script as a teleprompter while recording video is courtesy of a Google for Learning course.
Collaboration in Vids happens like a typical Google doc -- you can share viewing, commenting or editing rights to your project. If it's viewing only, the viewer will actually only be able to play the video from your project. This deserves a bit of emphasis, because in a practical sense it means your Vid becomes a real-time living artifact just like a Doc or a Sheet; you could share a link just once and the viewer will always see the most recent updates, as opposed to having to delete an old video and resharing a new link to the latest uploaded version. This alone can save significant time and increase workflow. That capability aside, you can also download the final video as a MP4.
You can create Vids as an assignment in Google Classroom for your students! First, create a template that you want students to customize and add their own text and materials. (One of the fun Inserts to consider is a "Video Placeholder," available in various cutout frames; when an editor comes across one of these icons, they can click on it to record their own video that will be automatically matted to the chosen shape.) You can then assign it in Classroom, making sure it "force copies" for each student. Students can open the assignment, edit it and turn in the Vid like any other Google doc. Here's a short video about creating Vids assignments in Classroom, from a free Google for Learning course:
There is admittedly a cookie-cutter aspect that could occur when students are creating videos within the templated perimeters of Vids. However, the intuitive ease of editing in Vids "in the cloud," as well as the ability to quickly share and collaborate, means that students should be able to spend more time on the content than on the clicking. The bottom line: you can make polished videos in a fraction of the time it took before, and for free.
It's worth noting that if you pay for an upgraded AI license, you can also use Gemini inside of Vids to create genAI video content or voiceovers. Here's a video (7:59) showing how Gemini integration with Vids works:
It used to be that improvements or new tools in Google were incremental -- you might be lucky to see something new every few years. As with all edtech nowadays, that speed of change has rapidly increased. By the time we hit spring 2026, I will expect new Google features that will likely make even the amazing ones in this blog entry seem quaint and "sooooo 2025."
It's another example of ChatGPT seemingly leapfrogging its competition. Since November 2022 when ChatGPT first broke through to the general public, other major generative AI platforms have come forward, but without as much fanfare. Microsoft's Copilot has an attractive user interface and, among other features, can generate accurate images inside of itself, allows voice-to-text input, and provides hyperlinks in responses, yet I hardly hear it as a person's first AI platform of choice. Google's AI went from Bard to the renamed branding of Gemini and was much like Copilot in its features. While Gemini and Copilot allows uploading of images, you can upload both text and image files to analyze or alter in ChatGPT. (A quick disclaimer: as you likely already know, AI is sometimes wrong and hallucinates. When it comes to inputs, I highly recommend trying prompts in multiple genAI platforms and comparing the results.)
But that was then, this is now. Gemini may have finally played its ace card with its latest upgrade, announced in a live webinar yesterday. (By registering, you can get access to the 45 minute archived recording.) Finally, Google seems poised to lean in on two of its major assets: its near-monopoly of education with its variously tiered domains, and having the world's most popular cloud-based productivity app suite.
Here are some of the highlights from the webinar. (Images are screenshots from the webinar unless otherwise noted.)
There is now a premium version of Gemini AI, one that can be incorporated across some of the core Google apps: Docs, Gmail, Slides, Sheets, and Meet. With the upgrade, AI help is just a button click away. This integration is likely a game-changer. Why pop open a new tab for another genAI when Gemini could be built right in?
However, for educational domain customers, Google recognizes there are special needs. Gemini will now offer, free of charge, several assurances: the data inputted from students and teachers will not be human reviewed, not be used to train AI models, and not be shared outside of their domain. (These three highly sought features are coming "soon," a phrase used several times in the webinar and in its various promotional material graphics.) For educators with privacy concerns, this alone might be a compelling reason for a school district to recommend Gemini over other genAI tools. Vivek Chachcha, Product Manager of Gemini Education, promises that the AI tool will help "save time," "make learning more personal," "inspire creativity," and help students "learn confidently" (by "empower[ing] students with guided support").
The webinar included several video examples of Gemini at work in various Google apps, and some shorter excepts of these videos are available separately on its Google for Education YouTube channel.
In this clip, an instructional coach uses Gemini to draft a professional development session agenda in Sheets (35 seconds):
In this last clip, a teacher inside of a Doc creates a first draft of feedback on a student's poem (47 seconds):
In a related tool revealed earlier this week, Google also has a "side panel" feature that will incorporate Gemini. This will basically allow the AI to look across your various apps and files in Google Drive in order to complete the task. (Currently, access to this side panel requires being enrolled in Google Workspace Labs, which will likely need the approval of your domain's admin.)
An example of Gemini's side panel, inside of my personal Google Drive.
Yes, yes, you may be saying, but what about the price?
First, it may be helpful to compare the current free Gemini chat AI (accessible in a separate tab, at its own URL) versus the new paid Gemini for Workspace:
Next, note that these paid licenses come at two different tiers of pricing. (The upper "Premium" license includes Gemini being able to have "advanced meetings" in Google Meet, a potential nod to the ChatGPT meeting summary/interaction feature mentioned at the start of this blog entry.) The good news: there is no minimum number of licenses you can purchase for users, and there is a discount if you order with a yearlong commitment before August 23, 2024. The bad news: these premium features do not come automatically with any current upgraded Education domain tier, although Education Plus customers can qualify for a bit more of a discount, again if ordered before August 23.
While the integration of Gemini across your Google apps does cost, it is easy to see how powerful the AI could be for Google Suite learners. If nothing else, it certainly gives Gemini an opportunity to jump to the front of the genAI competition line!
I'll end this blog entry with some Gemini resources:
For tech and IT friends, read this blog entry from "Workspace Updates" (5/23/24) for an official Google breakdown of Gemini for Google Workspace for Education's coming upgrades.
If you're interested in more of a general audience narrative of what's happening with Gemini in education, here's an entry from Google's blog on 5/16/24.
Looking for even more ways to use Gemini for Google Workspace in education, from Google itself? Check out these Slides from April 2024. The resources and tips include advice on writing better prompts, multiple visual examples of how to use Gemini in various apps, and help for domain admin.
Google offers some free online courses. Here's one I just completed myself, and I highly recommend for a low-stress walkthrough: "Get Started with Gemini for Google Workspace." (It requires logging in and allowing ExceedLMS to access your Google account, but again, it's free!)
One last thing: have a great summer and enjoy some time off!
