EMEA AI Afternoon in June

Great. Thank you so much. Thank you everyone for joining us this afternoon. I know two hours is quite a lot to take from your busy schedule, so we really appreciate it. We really wanted to take the opportunity today to spend some time with our customers in EMEA to talk a bit more about our Ignite AI developments, what's coming, what can you expect of all the different features we have, what are some practical use cases you might be thinking of? So we really wanna spend time on that today. We welcome your engagement.

There's gonna be lots of time for q and a, so please feel free to drop your feedback and thoughts in the chat. And if you have actual questions that you'd like us to address, please put them in the q and a section. My colleagues and myself will be looking after those questions during the session, and then we'll raise a couple during the q and a parts as well. So on the next slide, we've highlighted the agenda, but actually so sorry. Just to go back to it, I'll before we go into that, I'll introduce your speakers for the day.

So some of you will know me already. My name is Laura Slotmakers. I'm a customer success lead. I work directly with a couple of our customers in the Benelux region, and I also lead a small team of customer success managers who look after multiple countries in the rest of Europe, Middle East, and Africa. I'll be moderating us today, but your main speakers for today will be my colleagues.

We've got Jody here, who's our senior director of academic strategy and innovation. We've got Zach, our chief architect. And those of you who joined our webinar yesterday, which focused specifically on the EUAI act and how we're taking that into account, You'll have already heard Jody and Zach speak. They'll take some of the agenda today. And then we've got my colleague, Alex, joining us as well who's our principal solutions engineer.

So on the next slide, I'm going to take you through the agenda. Whilst I'm gonna go through that, I'm also going to quickly pop, a poll on the screen because I'm kind of curious to understand where everyone is in terms of current usage of Ignite AI tools. So feel free to have a look at the poll and fill that in. In the meantime, I'll run you through the agenda so you have an idea when to take your breaks, and what to definitely stick around for. We'll kick off today with Jody and Zach talking you through Ignite AI, what's Instructure's vision, how are we developing.

That will probably last around thirty minutes. Then we've got ten minutes to talk specifically around Canvas tiers. I think most of you will already have heard about Canvas tiers, and, obviously, as a whole, that covers much more than AI. But we wanted to make sure that you understood the tiers because it is quite important context to know what you currently have access to and what you'll have access to long term. So it's really important context for you to have.

We wanted to be quite transparent on that as well. We'll spend around ten minutes on that, and that should leave us quite a lot of time to do an initial q and a. We'll do that q and a similarly or simultaneously to a break. Feel free to grab coffee, grab tea, leave your screen if you need to, but we'll also use that time to raise some questions live and get some engagement going. And then the second hour, we're gonna go much more in-depth about all of the individual feature options that we have, why might you use them, what are outcomes that you could achieve with those, and we'll be demoing them as well.

So that will be mostly my colleague, Alex, doing that. And then at the end, you'll notice on the screen that we will still have some time left, so we'll have more time for actual discussion and q and a at the end as well. We really wanna make it as engaged as possible, so please interact where you have questions or where you have feedback for us. So that was the agenda. Before we actually move on, I'm gonna have a quick look at the results of the poll, and I'm gonna end it here.

So it seems like, actually, most of you at the moment, more than fifty percent are not using multiple Ignite AI tools. So that might mean you're using one. It might mean you're not using any at all. But there are actually a a good percentage of you that actually do use Ignite AI, about thirty six percent of you. So I'm quite keen to understand what might be holding some of you back.

I see some of you have mentioned this might be due to institutional policy, so we really hope that yesterday's session as well as today's session might alleviate some of those concerns and might put you in a position to confidently start using those tools. That's really it. We'll move on, and I'll hand over to Jody and Zach. Fantastic. Thank you so much, Laura, and thank you, to all of you who are joining us this afternoon.

When I joined Instructure fourteen years ago, we were having a lot of conversations about a different new technology, and that technology was the cloud. We were the first learning management system to deploy natively into the cloud, and we chose that technology not because it was new or it was popular or it was exciting, but because it it delivered things that we knew educators and students needed. Things like better reliability, better availability, and better user experience. And so now as we look at AI and what that means in Canvas and what that means at your schools, we're taking those same lessons that we learned, in deploying the cloud, and we're applying those to our AI strategy. And as you'll see today, that means that we're focused first on the classroom and on improving teaching and learning and not just on the tool that we're using.

But, you know, Jody will will tell you here that we also know that technology is not the only challenge, and it's not the only solution to what you're you're dealing with. Next slide. Yeah. That's right, Zach. So before we get into what we've built and exactly how we're building, we wanted to start with where what we're hearing from institutions like yours.

So in a recent study, we asked educators and educational leaders where they were struggling, specifically around AI and introducing AI to their classrooms. And, really, a clear picture came back, and we continue to hear these still. The top challenges really aren't, like Zach mentioned, about the technology themselves. They're about the trust in the technology, the people, and the processes. So more than seventy percent told us their biggest worry is with fairness, ethics, and institutional risk.

Nearly sixty percent are concerned about accuracy and reliability. And, of course, just over half point to faculty buy in and adoption. There's a lot to consider as we think about adopting different tools into our classrooms. When we asked where they wanted help, the answers lined up. Around two thirds wanted help framing AI to assist educators, not to replace them.

Such a critical component, of course. A similar number want help vetting and putting AI to work well in their classrooms and across their institutions, and nearly sixty percent want clarity on where the human oversight actually fits into this this tool and how to use it. So here's the takeaway. Great technology on its own doesn't solve this, as Zach mentioned. AI is moving very fast, and there are so many tools that are free for anyone to pick up and try.

And that might be easy for one educator or academic, but scaling it well across the whole institution safely and consistently is definitely a much harder job. And that's really the job that we want to help you with. So as we think about the architecture on the next slide, that's exactly why we are building and have built Ignite AI. We want Ignite AI to be a secure in context AI for education. It helps in two different ways.

First, native end product solutions that you will find and that you'll hear a little bit about later with Alex. You get ready to go AI solutions right inside of Canvas where you're already working. So there's less setup for you and your team and more value straight away. The second is, with orchestrated AI ecosystems. Just as Zach mentioned, we have been that open platform, and we want to continue to be.

Ignite and Ignite AI really connects us with other tools you already use. We've, continued to try to be that ecosystem across time and, again, want to continue to bring AI into that open space environment. So the word I keep coming back to as I think about this is a conductor. Ignite AI sits really at the center and brings everything into tune. So rather than handing you one more instrument to play on your own, we really want to create that ecosystem that allows you to bring those systems together.

And to share a little bit more about that, I'm gonna actually hand it back over to Zach to show you a little bit how this works under the hood. Thanks so much, Jody. So at the base of Ignite AI, are you will find the investments in our open platform that we've made over the last decade and and beyond. And that starts with all of the APIs that we have available inside of Canvas. In fact, over five hundred of them, which mean that if you can do something in Canvas as a human, it's possible to write a program that does that same thing for you or can automate that or simplify a a key workflow.

Now, those have always been a really important part of Canvas' capabilities, but they're especially important as we talk about artificial intelligence, as we talk about things like agents, because these mean that those tools now can integrate more deeply with Canvas, in a way that, as as Jody mentioned, feels very native, is is easy to to understand and and easy to adopt. And so, on top of these APIs, you see us now making additional investments, to make sure that it's, as easy as possible for us to build AI features inside of Canvas, but also easy for you and for your other partners to integrate with Canvas as well. So on the next slide, we'll see that that's not just about the APIs. That's also about things like LTI one point three, making sure that we have great support there and that we have a number of of placements, to make these experiences embed directly into Canvas. And then it's also looking at AI specific tools, things like model context protocol servers, which provide a standard way for, large language models to understand what APIs are available and to call them.

So that, again, whether we're building the feature, your partners are building the feature, or you're building the feature, you can know that it's going to look and feel native inside of Canvas and integrate better there than it would anywhere else. Slide, please. And all of that sounds great, but it only works if you understand what it is, where it's hosted, and have confidence in your ability to use it. And so we wrap all of our AI features inside of Canvas and Ignite with what we call our nutrition facts. These are small cards that will give you the information you need to decide if an AI feature is right for you and for your institution.

So that includes things like, where the model is hosted, what model it is, what data is being used by the model and how, and what the expected outcomes and the expected risks are. Right? So something like translation feature, for instance, may make a lot of sense in one class but be a very bad fit in a language class. Nutrition facts help you identify those types of challenges and, again, allow you to customize Canvas in a way that makes sense for you. On the next slide. And you can do that through all of the standard feature flags that you're used to in Canvas today.

So all Ignite AI features, as I'm sure you're aware, are off by default, but you have a lot of control and flexibility, in how you roll these out, whether it's at for the entire institution, for a single subaccount, for or even to a single course. Next slide. And I'll hand it back to Jody. Thanks, Zach. So let's talk about why.

What does Ignite AI actually do for those of you that might be using it every day or as you consider to think about using it every day? The first way that we were thinking about this was really about driving educator efficiencies. We know that the job of the educator is difficult. It is time consuming, and there is a lot that you have to do, a lot of decisions that you're making, and a lot of things that you're building. So, really, we wanted to have Ignite AI streamline some of those processes. In this example you're seeing right now, you see that it is actually helping to streamline the process of building a rubric.

As you listen to Alex later this afternoon, you'll hear about many other ways, excuse me, that we're building in those efficiencies. Excuse me. A job that used to eat up an afternoon now can take just a few minutes, and that time goes straight back to the work that you're doing as an educator to help make sure that you are meeting the needs of your students. And those are the jobs that you only you can do. Right? So knowing your students, shaping a lesson, having the conversations that really change how someone might learn.

We want to provide opportunities for you to use Ignite AI to handle those repetitive first drafts so that then you can, make those final decisions, be the judgment, be the context, the final word, and give you more time back to actually be working with your students. In addition, on the next slide, you'll see that efficiency is only half of the story. The other half is what this does for students. Here, you can see that Ignite AI is supporting learners directly, helping to provide some meaningful feedback to help students actually to stay on target in that in that quick video there. Same platform, same guardrails, but now the value reaches the student, not just the educator, and that's the point.

We want this to lift the whole experience on both sides of the classroom. And then lastly, here's what makes all of this stick on the next slide. Ignite AI isn't a separate product that you bolt on like we mentioned. It lives inside the workflows your educators are already using, so there's almost nothing new to learn. A few things make it easy to trust and easy to adopt.

You'll see consistent indicators wherever AI is at work. So you'll see this this clear sparkle, as we call it, icon and some tool tips. So everyone knows when AI is in play and choose when to use it. The features show up in line, as Zach mentioned, right where the work already happens. And for the bigger jobs, an educator can hand a multistep task to the Ignite agent, which you'll hear a little bit more about later, and let it work across our APIs, even reaching other a AI tools in your ecosystem similar to the way that Zach was just describing all from a single prompt.

So adoption isn't really a project that you have to launch. It's a set of of helpful options that appear and that flow within the work that you're already doing on your time on your terms. And then, through the way that we have architected that Zach just shared, not only the interactions, but the oversight for what, when, and where to enable AI features is up to you also or your administrator at your institutions. So Zach and I have shared the what and the why to show you how this comes together in practice and what it looks like across our Canvas tiers. But I believe next, I'm actually handing it then over to Alex to talk a little bit about those tiers.

Yeah. Perfect. Thanks, Jody. So Laura mentioned at the start during the the agenda that, we would be touching on Canvas tiers today. You know, this this spans more than just the AI features and functionality that we'll be talking about today and what Jody and Zach have already discussed, but it is an important part of the story because it's it's a change that allows us to focus the way that we approach the development of the platform going forward.

Right? So before we get into the platform, and we'll we'll see all of these different features, everything live, and seeing where the value actually comes from, let's start by just breaking down for for everyone on the call today what this move to Canvas tiers actually means for everyone. Right? So what we're doing is we're transitioning from a a fragmented sort of, let's say, patchwork type contract situation. Right? And we're moving to a tiered structure. So that's what we see on the left. We have, you know, the ability to have Canvas LMS.

And, historically, you could add on, you know, specific pieces like the agents or, ask your data with Intelligent Insights or Canvas Studio. So moving to the right where we have, these three tiers, this is going to allow us to better align things like the features, the AI capabilities, and the value that you at your institutions will be getting out of the relationship within structure into these three tiers. Right? So if we can go to the next slide, we'll we'll sort of zoom in on each one of these. So the first one that we see is Canvas core. Right? The next level up is Canvas plus, and then the third tier is Canvas next.

So if we go to the next slide, we'll see actually what is included within these. Right? This is this is broadly what we're looking at from a from a tiering perspective. Right? Where Canvas core is the same, reliable experience that you have today. Canvas plus is looking at leveling that up, looking at making learning visible, engagement actionable, and and an easier entry point to high quality teaching. Right? And then finally, Canvas Next is, for those institutions who really want to, you know, push that extra mile and and move into those transformative workflows.

Right? So if we go one more slide here, we'll be able to see exactly what we'll focus on, later on here when when we start getting into, the platform itself. I've pulled out and extracted the specific AI functionality that is, inherent within each of these tiers. Right? So within Canvas core, there are a number of of areas of the platform that leverage AI, and we've kind of bundled these together as what we call Ignite AI Essentials. So we'll see those. Once we we move into that level up perspective, we'll start looking at some of these tools that are specific to to teaching and feedback.

Right? And then finally, in Canvas Next, that's where we'll see those tools like the Ignite AI agents, the ask your data tool, and the Ignite AI study tools as well, which will be more from the student's perspective. Right? Thank you, Alex. Thank you, Jody and Zach. I think we're a little bit ahead of schedule, but that's great. That means we have more time for interactions and engagement.

So we'll take about twenty minutes now for q and a or for a break if you need it. So feel free to drop questions in the q and a section. I'm happy to raise them out loud and live and address them there. If you wanna share some feedback or share how you've currently been utilizing Ignite AI, we'd love to hear it. So feel free to drop that in the chat as well.

And like I said, we'll be using the next twenty minutes to do so. So we'll go back into the next part of the presentation, which is gonna go much more in-depth on each individual feature, why you would be using them, what outcomes you can achieve with them, and we'll start that at, I guess, depending on where you are, either at two forty five or if you're in Central Europe at three forty five. Again, any questions that you want to ask, feel free to put those into q and a. Can see a few of those going in already, and we can, specifically focus those questions on some of the things that Jody and Zach have have spoken about today as well as Canvas tiers. And, again, there will be another q and a session once we've had a chance to run through the platform and see some of these things in action.

Exactly. Yeah. That's a good point, Alex. There might be many more questions that you can't think of yet, but that will undoubtedly come up in the second half. But we've got a question from Jeff.

What's the difference in LLM models in the different Canvas tiers? I would say this is one for Zach. I can see you unmute yourself. Thank you, Zach. Yeah. Very good question.

So we don't choose models per tier. We choose them per feature. So, you know, a feature in Canvas core may use a large model if that's appropriate for that feature. But most of our features are powered by anthropic models, so some version of either Haiku or Sonnet. Thank you for that, Zach.

We've got a question from Ismail. I suppose that the different tiers have different pricing. And thank you for asking the question because I'm sure that's what a lot of us or a lot of people on the call are wondering, and you're absolutely right. The different Canvas tiers will have different pricing. So we'll announce those pricing models soon, and, obviously, you can work with your institution CSM to find out what that might mean for your institution if you're wanting to move to Canvas plus or Canvas next.

Then we've got a question from Beata, and I do apologize if I'm pronouncing your names wrong here. The biggest question is about GDPR. What data is gathered, saved, where, and for how long? Is there any risk that any student or teacher data is in danger? Which I'm sure is a very timely question as well that I'll hand over to Jody and Zach to to address. Yeah. So we I guess storage and and logging, again, depends feature by feature on on what we do.

We do publish that information in the nutrition facts. So I think that's, one of those key pieces of information you can use. I I will say, I think importantly, we do not train any of our AI features or models on your institution data or on student data. Now, as far as the the question of of data leaking out or, you know, becoming available to to someone, I I think I I have learned to never say never, but I think the chances are very low, because, all of our features, while we're using anthropic models, we host those inside of AWS. So all of those AI capabilities, are are hosted right next to the rest of your core Canvas infrastructure.

So we're we're not sending that data out to a third party or or out onto a a public, network during transmission. Thank you, Zach. And Berth also clarified she's mainly or they're mainly interested in the Ignite agent when it comes to these questions. Perfect. Okay.

Yeah. So Ignite agent, I I believe okay. I was I was going to say how long we log for, but I'm gonna get it wrong. It is in the nutrition pack. So Yeah.

Thank you, Zach. We've got lots of questions coming in, so I'll move forward. Does Ignite AI process data outside of the EU for EU based users? I believe it's already been answered, but it's worth clarifying it specifically. Right. It it does not.

No. All all data remains in the EU. Thank you, Zach. So there's a couple of questions coming from Adam. Have we completed I think this is an interesting one, and I'm sure there's gonna be more institutions dealing with it.

Have we completed any analysis of the environmental impact of the AI models or tools that we are bringing in? And the reason, they're asking is because their students are very conscious of it and are actually against the use of AI because of that reason. Yeah. I I love this question. So, there, okay. There are some things we know.

There are some things that we we don't that are just difficult for us to to properly estimate. Most of let's see. How do I where do I wanna start? You you've asked a question. It's gonna take me two hours now to answer. I love this topic.

So I I'll start, with Amazon. I think we have a very good partner, in Amazon. They're committed to sustainability. Most of their data center power comes from renewable, sources like solar, where they do not have access to renewables. They, they purchase offset credits, which I mean, they're they're not perfect, but it's it's something.

And they're also committed to being water positive everywhere they have a data center by twenty thirty. So they are, they're well on their way there to actually providing more water for the communities, that they're hosted in than than that they use. As far as the the exact, water use, carbon offset or carbon generation and electricity used, for our features, we don't have, that information. We do base our decisions on industry standard and and typical reports. So there, Google recently published a paper, that described the, the energy and environmental impact of their AI features.

I think that's fairly representative of of what these things cost across the board. Now that paper suggests that, a single prompt is about as expensive as, running your microwave for one second, driving your car for twelve centimeters, and then the the water usage is it's about five drops of water. Now the comparison I like to use there is that, building a t shirt, uses enough water, to, equate to one point three million prompts. Or a smartphone, uses enough water, during its creation, to equate to about six point five million prompts. So, when you use Ignite AI features, those are the things I I think of.

Right? So if I'm I'm summarizing a discussion, it's about on par with running a microwave for one second. Now I think our features are a little more I I guess a a little less expensive than even that because we tend to use very small models. And that's where the nutrition facts can help a little bit. Something like Haiku is on the very low end of of consumption. And and we do that on purpose.

Right? That's that's part of our process because we do want to minimize the environmental impact of these wherever we can. Thank you, Zach. Let's move on because there's quite a few more questions. I know there was another question about the environmental impact and that we noticed or they noticed it's not in the nutrition facts. So I think we've already addressed how we are looking at the challenge, but it's it's good feedback about the nutrition facts.

And we can certainly take that in words and see if that's something we can we can add there as well. So thank you, Graham. Then we've got Jordi. Jordi's asking whether our solution will and this is a really interesting one. I knew this would come up as well.

We'll support the integration of custom AI models. For example, models developed or trained by an institution, particularly in relation to MCP servers or similar mechanisms for connecting external AI services and tools? Yeah. So we the answer here is yes and no. Very good question. We, through MCP, it is possible to take any model that you have, and connect that direct to Canvas APIs and have it do things for you in Canvas.

You can have that run on its own, or you can use LTI to embed that directly inside of Canvas. So that's the yes. The the no is that you cannot then take that model and use it to power the Ignite AI features that we've built, those inline features. So I I know you mentioned rubric creation, for instance. You can't use the Ignite AI, rubric interface with a custom model.

Right? That that's locked to the model that it was built with, But you could go build or connect your own model to Canvas and then have that model create rubrics for you. Thank you so much, Zach. I've got a couple more questions coming in. Some of them are related to the tiers, so there might be questions that for example, Julie, I'm reading your questions about the packages, and it might include features that you're not necessarily interested in. So I think that's probably worth a specific conversation with your customer success manager because it will look quite different for you than it might look for others.

But overall, the reason we've packaged it all this way is because we really want to simplify our processes. And we want to change our our way of looking at Canvas tiers. Whilst we might have worked with a lot of add ons in the past and all had different names and and, use cases, it does make sense to consolidate them. And similar to how when you're using Canvas core, you might be focusing your usage of certain features that exist in Canvas, you might not be using others. It'll be a similar approach for the plus and next year.

What will be really important for everyone here is to, once you have the clarity of what the model will look like and what the pricing might look like, to make that investigation internally, whether what you're actually getting out of it is worth the investment for you. And whether that means you're getting additional features that you're not actually going to use would then hopefully no longer be the focus of the conversation. So that's a very generic answer, Julia, but I recommend that you take that up with your customer success manager to have that specific conversation. Paul is asking if we can have a link to the statistics around AI and the environment. So, Zach, you mentioned a couple of things related to AWS and some research that's done.

Is that something that we might be able to to share publicly, Zach? Yeah. Absolutely. We can put that together and share it. Yeah. So what I would recommend doing here for the audience, I'll speak to all the speakers here afterwards.

We'll collect some of the resources and make sure they get shared afterwards as well when we reach out with the follow-up, conversation. Let me have a look at the last question. Does any of the tiers allow users to search and scan via OCR through documents such as docs, PPTX, or PDF within the Canvas files in the instance that either a summary is requested or a complex question is raised? I'll have to admit I do not know the answer to that question. Is it something, Alex or Zach, you know, or shall we park it and answer it later? So not today. I think that's something that we you're not the only person, who has given us that feedback, so it's something we're we're looking at how we could do.

Thank you, Zach. What tools will we be including in the core product to ensure data stored within Canvas can be better managed to meet GDPR requirements, especially when AI is going to be getting access to data that previously may have been considered, stale or no longer needed? This is a question from Tim. This is another one where I would say, stay tuned. We we are working on some updates here. Yes.

Exactly. So we'll we'll be communicating once we have some more clarity for that. Thank you, Tim, for raising. And then many institutions deploy external content on Canvas courses, not native Canvas content. Will Ignite be able to read and use that content? So today, the answer is usually no.

It is something we're having a lot of discussions with, or or discussions about with our partners, because in that case, we need their help as well, to build the integrations that we use. Yep. So I guess it ultimately will depend on those partner relationships. Thank you, Zach. And then currently, the last question we have in the q and a, Ignite AI users cannot see the history of the use of Ignite AI rights.

I guess, just to be sure, we need to clarify that. Yeah. That's that's correct. Thank you. Then we've got a new question coming in from Jordi.

Is it possible to have different licenses from different tiers within the same instance? Okay. So when we're looking at the tiers and all the individual potential setups, this is really something we'll have to individually follow-up on because it really depends on what your current contract looks like, what your user base looks like. So, Jordi, that's something that I would once again recommend you to reach out to your CSM for, and we can make sure that you get the questions you need or the answers you need. Sorry. Right.

The q and a has gotten quiet now. I think I mentioned we would start again at two forty five. So to make sure that people know when to come back, I'll leave five more minutes of q and a and break time. So we'll stay here. Feel free to ask questions, but we'll still or we won't continue for the next five minutes.

I see we've got some more questions that reached me. So we've got a question. What does Canvas tiers mean for my existing contract? Now once again, as I've mentioned before, when it comes to your contract specifics, this is really something to reach out to your customer success manager for, whether you're already using some of the functionalities that are on the higher tiers. It'll look different for everyone. So please, reach out to your customer success manager, and they'll get you the answers that you need.

Yeah. There will definitely be some nuances based on your institution, but it might be worth calling out here as well that, there's no there's no uplift to get into Canvas core. Canvas core is the experience that you that you already have today. Right? So there will be nuances. If you are using something like Canvas Studio and and some other tools that you you might already be falling into that Canvas plus territory.

So that's where you need to have those individual conversations. But, yeah, there's no, there there's no, you know, pace of play for Canvas core right now. That is that's where you exist. That's that's where you've already been going through that, you know, experience of reliable and accessible learning. Thank you for clarifying that, Alex.

I just want to point out for those who haven't seen it yet. I put some resources in the chat. So when you go back to the chat, you'll find all the nutrition facts that Zach mentioned earlier, and you'll also find a feature comparison table between the different tiers. It'll include the same information that Alex already shared, but then you'll have it, in front of you. So no need to to make notes here.

You'll have it all available on the community. It's about quarter two now, so let's continue with the rest of the session. Like we mentioned before the break, the next hour of the session is going to go much more in-depth about what we have available in terms of AI feature options, why you might be using them, and we can actually show what they look like. So I'm gonna hand over, to Alex to run you through that. And then once again, we'll we're way ahead of schedule, so we'll have some time for deeper q and a at the end if further questions come up.

So please keep putting them in the q and a section. Perfect. So I've I've put the tiers back up on screen here because I do want us to anchor onto these tiers as as we move through over the next thirty to forty minutes or so. The idea here is that I want to make sure that everyone comes away from this session with an understanding of what you have access today, so things that are already live in in your platforms or things that can be switched on. Right? You have the autonomy to go and switch on.

We know that, you know, lots of institutions, even from that poll at the start of the conversation, we know that lots of institutions, for various reasons, haven't turned on some of these features yet. So it's today's a look at where that value comes from for each of those individual tools and and features that that we've, woven into the platform. Right? So that's the focus today. We're gonna walk through, you know, for each tier, the the AI functionality, and we'll see exactly the types of problems that are being solved and the value that the individuals at your institution will be able to derive from those. Right? The value your staff can can get from the tools that we'll look at today, the value your students will be able to get, and how that affects the actual experience, right, the day to day experience of education.

So we're gonna start here with Canvas core. Right? And I'm I'm using, sort of our, our line here to to anchor each part of the conversation. Right? So Canvas core, if we boil it down, it it's thinking about, again, providing that reliable place to learn, while having an accessible experience. Right? So this is where we're going to start. Canvas Core is that foundational layer.

Right? The we're going to be looking at features that are embedded into the everyday teaching and learning experience. So we'll work our way through, and I'll introduce each of these features that that we've woven into the platform. Then I will introduce a persona that we'll see the platform from, that we'll see it from their perspective, and then we'll jump into the platform. And we'll follow that cadence or track as we go through Canvas core, Canvas plus, and Canvas next. Right? So some of these will be familiar to you, and some of them may be completely new.

Right? You may everyone will be at different journeys in terms of, how much of their own research, yeah, you've done or or, you know, sessions that you've joined with your CSMs. But I just want to make sure that we have a level playing field here and everyone understands what's what is involved here and what's available. So remember, Canvas Core, this is what you have available to you today. So the first piece is the question authoring tool for quizzes. Right? So the idea here is that you'll be able to streamline assessment creation, actually generating questions in a quiz, but not just with, you know, zero context or zero source.

You can utilize the the, content within your courses to use as a source material and build your questions off of that. Right? So that's the first piece that we'll be seeing. The next is Ignite AI search. Right? So this is all about, you know, leaning into that accessibility side of things. It's it's making sure that students can actually access the content, quickly and easily.

Right? Being able to get to the exact piece of content or assessment or discussion that they might be trying to find, especially in in some larger courses, and giving them that access at their fingertips. Right? Keeping with that accessibility context here is the content accessibility checker. Right? This is a fairly new one. This is a bit newer than the other ones that we've just seen, and it is a a tool that we've kind of expanded within the context of a Canvas course. Right? In building content in Canvas, many of you will be familiar with the accessibility checker that exists within the rich content editor.

Everywhere you're building content, you can check that individual piece of content or accessibility standards. Right? What we've done is we've leveled this up while still keeping it within that Canvas core tier. Right? So we'll be able to actually surface accessibility issues across your course all in one place. So as a as an an educator, you have the ability to see that in one place. And, also, where it's relevant, utilize artificial intelligence to take some of the take some of the lift of remediation off of the educator sense.

Right? So specifically in things like, alt text that might be missing or is just the image file name, table headers as well, you'll be able to actually use the use Ignite AI to, give a first pass at what that might look like. Right? Finally, here we have, two more two more pieces, which start moving more into the the collaborative perspective within the within the platform. And that is, first of all, the discussion summaries. Right? So as an educator, being able to identify the topics, any themes that might be appearing within larger discussion forums. I always really like to to focus on the use case of, bringing topics that might need further attention in a in a live session or or, you know, potentially a face to face session.

So it's quite a few different use cases for the discussion summaries tool, but it's all about allowing educators to refocus efforts, right, to make sure that the conversation is as tailored as possible. And then finally, keeping with that, that theme of of communication, it's the translations functionality. So for discussions, announcements, the inbox, the ability to translate content in real time and also directly within the platform. Right? Traditionally, you would you know, if you receive something that you weren't quite sure because it was in a language that is not your first language, right, you would take that, you would copy it, and you would go and paste it somewhere else in in Google Translate very traditionally, or, you know, today, you might put it into another large language model. But the beauty here is that you can keep you can keep everything within the context of Canvas, right, without having to leave and go somewhere else.

You can translate content into a language that you're most comfortable with so you can break down those barriers to communication. Right? So the those are five different tools or features here that we're going to cover, live within within the platform. And to do that, I'm going to introduce Sarah here. So Sarah is a lecturer, at a midsized European university. And like many of your lecturers at your institutions, I would imagine, has a lot on her plate.

Right? There's a new intake that's starting, multiple courses that need updating. There's a discussion forum that's been running for a while. And because they have, she has some Erasmus students as part of her courses, there's different languages that are being used in the discussion forum. So there's a lot to get through, and Sarah just needs to get a bit more time back on her plate. Right? So let's let's have a look and see how Canvas Core in particular right? We're just focused on Canvas Core right now.

We'll see how we can affect Sarah's day. Right? So let's jump into the platform here. Right? So the first thing that we're going to take a look at here, being logged in as Sarah, is how we can go and build some assessments faster. Remember, I I have a new intake of of students. I need to make sure that they have the content and and assessment in front of them that they'll need to be going through.

So I'll come and have a look at my module one knowledge check here. Right. So in the in the quiz builder, with the question authoring or item authoring feature switched on, what we'd what we'll be able to see is all of all of the typical tools that you have at your fingertips, creating things like multiple choice and, you know, matching and ordering and so on. But alongside that, we have this generate with I with AI button. Right? So it's something I think that, you know, Jody and Zach mentioned a little bit earlier that some of the beauty behind, you know, Ignite AI as a platform as it's woven into Canvas is that you're not having to go to external tools and go and find some other context somewhere else to to leverage the power of AI.

And not only that, as a teacher, I don't have to learn a brand new tool. This context is all sort of the same user interface that I'm familiar with. Right? The only thing I have to do is just select a new option for creating content to generate with AI here. Right? So what I'll be able to do is, of course, define some context for the questions that I want to have asked. Right? So that comes from this source material section here.

And you'll notice there's tooltips throughout the the features that leverage AI just to make sure we're being completely transparent on on what is happening and also how best to use the tool. Right? So just to make a note here, you will always be able to get straight to the nutrition facts directly within the platform. Laura shared a link to a a a public facing version of the nutrition facts. But I can see here, for example, the base models model that's being used, whether it's trained with user data, which is an emphatic no here, as well as some other pieces here, right, all within the context of the tool itself. So let's go and select some content.

I'm going to select just my core content here. Right? This is my module one knowledge check, so I'll use the core content, and I can use, you know, this paper from Lund University as well. So I have a couple of pieces of source material. Right? This is where everything's going to be anchoring off of. Now I can also add some text.

I can upload a file if I wanted to if it wasn't already in context within the course, but I have everything I need there. I can further tweak what that focus will be. Right? If this paper by Lund University was focusing on a number of different topics, I could actually focus that in with my text box here. And then finally, I can get really into the into the details of the pedagogical approach for this assessment. Right? I can decide what learning outcome.

Maybe I wanna use this, learning outcome for identifying and explaining foundational theories. I can define the depth of knowledge that I want here, so maybe just recall and reproduction, keeping things simple. This is module one knowledge check after all. And Bloom's taxonomy, let's say, let's say comprehension here. Right? Now I'll I'll create three questions here just to give us a a bit of a some context here.

For now, by the way, this is limited to multiple choice. As we move forward and continue developing, this is this is a drop down for a reason. Right? This this will expand, as we move forward. So what's happening here, and this is important to note, that this is going to create a structured draft of assessment items. Right? As the teacher, I still have I'm going to be able to make every call on what is going to go in front of my students.

Right? So the difference between those two here, that's you know, I'm I'm editing questions rather than starting from scratch. Right? So I'm I'm remember, Sarah needs some time back on her plate. This is one place where I can get some time back. Rather than building these questions from scratch, I can go in and edit them as I move forward. Right? I can make any changes to this first question, right, whether that's just ordering, selecting which is the specific point here.

I can, make changes to the individual points as well, and then I can move forward. Right? Let's say that all three of these look good to me. I can go ahead and add those to the quiz. So what could have taken quite a bit more time, right, for Sarah to go and build these questions, it probably would have leveraged another large language model as well potentially. Right? Something like ChatGPT or or Claude or, you know, or Gemini.

We're able to keep that experience centralized within Canvas, within the context of Canvas, and the transparency that comes along with that. Right. So once Sarah has built some some assessments, has built the content, right, everything's where it needs to be, the next piece is making sure that the courses are actually going to work for every student. Right? And that is where we come into the topic of accessibility and and sort of a smaller part of that word just access. Right? Accessing the content.

Sorry, Alex. Could I just stop you for a second? Because we've got one question in the q and a that's very specific to what you just showed, so I just wanted to make sure to raise it now. Jeff is asking, and this was about I believe it was about the quiz generator. It includes only one extra file. Will this be expanded? The PowerPoints and other link files in the course are not searched, or are they? That is a good question.

That might be one that we need to pin for now. What the what the content comes from like, actually, if I just jump back to this here. When we're looking at content here, this is coming from the actual module content itself rather than, you know, the the course file directory. So it would need to be in the context of a module already. But to the point around the one extra file versus and and the sizes of those, that's something that we can make a note to follow-up with around the specifics and and potentially what the development might look like as well.

Yes. Exactly. That'll be a question we'll raise with with our product development team, Jeff. But I think when you mentioned the linked files, as Alex has shown, he actually chose a linked file as one of the resources for the questions that he built. Thank you, Alex.

I'm sorry for, stopping you. We can we can continue on now. That is alright. That is what we want from this session. Right? We want this to be collaborative.

Even though there are almost a hundred of us on here, you know, let's raise the questions and make sure that we're all coming out of this session with a better understanding of what's available. Right? So let's let's jump back to the Ignite AI search. So the search tool is all about helping students find what they're looking for. Right? And and educators to an extent as well. But, really, this is about making sure that, you know, remember, I'm an educator here.

I want my students to be able to find the the content that they're looking for. Now part of that will be based on the way that I've organized my modules, but sometimes students just want to get straight to the point. Right? Search for flooding risk and see what's available to me. Right? So there's some core content that was that's returned, some some other pages of content down here as well, and I can filter any of this content. Right? So by different sources, whether that is assignments, discussions, you know, and so on, I can filter that down, and I can get right to the content that I might be looking for, including being able to jump straight there with a click of a button.

Right? Now this is important from a student perspective, but this is also, sort of inherently important for teachers as well. Right? If students can find what they need without having to raise a support query, right, that will reduce any drains on staff time that might be there today. Right? And it might not be massive in all institutions, but I can almost guarantee there are cases where students are having to to go and ask the question around where is where is this very specific piece of content. And every time that message comes up, that takes time for a teacher to respond to that, guide them to the right place. So being able to cut down on that at least at, you know, a certain percentage is helpful in the long run.

So, again, keeping on that that topic of access and accessibility, let's let's jump into that dashboard that I mentioned, right, the content accessibility checker here or course accessibility checker, I should say. So where in the past, I would go to a piece of content and I would make sure that that individual page, for example, had I had checked all of the accessibility metrics and made sure that we're where we need to be. It's time consuming to do that on a page by page page basis and a discussion and an assignment and so on. Right? As we add all of these up, it takes time. So now we have the ability to surface all of that in one place.

Right? We can surface any of those common accessibility issues where I can see there's there's sixteen here. Right? Sixteen are open. I resolved one already. And I can dive into these and fix them directly from here. Right? Again, from an efficiency standpoint, this is really where you can probably see the difference between jumping from page to page to discussion to assignment and finding the the things that need to be fixed versus having a list of sixteen things that I can fix all from one place.

Right? Let's look at this core content for sociological frameworks. So in this piece of content, we have an image, and this has been flagged up because the alt text is just the image file name. Right? We know that that's an issue, an accessibility issue from a screen reader standpoint, that this is not going to give any pertinent information about the about the image. So what I can do is I can let Ignite AI take a first pass at remediating that issue. Right? So now instead of being the image file name like it was before, now it is a protest sign reading the climate is changing, so should we.

Right? So this is exactly what I would probably put here from an alt text perspective. But instead of me going and typing that out, I've let the tool do that for me, and I can, make that fix directly from here and move on. Right? So this is one example. Right? I had one image that I needed to update the alt text on. Extrapolate that up, right, and start thinking about maybe you have ten, twenty different images that you just put in with the file name.

Now you can let this tool do that for you. And what could have taken, you know, upwards of an hour going in, typing out each one, saving it moving forward, could probably take less than ten minutes. Right? Alex, sorry. I have a question here for Magnus. Can you Fix or, you know, fix through the method you just showed, uploaded PDFs or PowerPoint presentation? So files that come outside of Canvas and that are attached in courses.

That's a good question, and it kind of relates to the other question that's in there as well. Designate search Yes. Support searching within files. So let's let's think about both of those, for a moment here. So right now, the the answer is that these tools are focused on the content that exists within Canvas itself.

Right? So content that is natively built within Canvas because that's where we have full visibility and control over accessing sort of the minute details of that content. So when we're checking accessibility, when we're checking, and searching and indexing, that's where we focused our efforts, at this point. Zach, I don't know if you have anything that you wanted to to speak to around direction of travel, for those pieces. Yeah. I I don't have anything right now.

I mean, you're right. This is, this is a a question we get quite a bit. I think that, for search, I I know I've had a number of discussions with standards groups to to make that, an education technology standard so that everyone could, can operate, but we don't have any news right now. Thanks. Cool.

So the final pieces here then that we'll look at from the Canvas core perspective, if you remember back, were the discussion summaries and translation. So starting to think more about communication at scale, and also across languages, right, not not just in a single language. As we, especially start looking at more global approaches to education, that translation piece becomes pretty critical. Right? So let me jump into a discussion forum here. Right? So this is, exploring foundations of artificial intelligence, bit of a discussion prompt, and I have a number of responses here.

Right? Now what I'll do to begin with, right, as a first pass is I can use the summary tool to give me an initial clear read on where the cohort is thinking. Right? This is especially useful for larger courses, right, where you might have upwards of a hundred responses to a discussion forum or just general collaboration, right, whether that's response or actually communicating with the with with peers, right, peer to peer communication. What I'm able to do is just at a broad level, get an understanding of what is being discussed, some things that are being highlighted, and maybe focus in some of these areas. Right? So what you see the summary is showing me is, okay, general good introductory understanding. Right? Some of the origins, some examples are mentioned.

There's some confusion around different types of AI. Right? Overall, introductory level grasp. Okay. So that gives me a good read so far around, like, what the discussion has been, whether or not this is hitting the level that I'm expecting at this point. Right? I'm very early in this course, let's say.

So an introductory level grasp is right on the money. That's where I want to be. Right? But now let's think about preparing for a tutoring session that's coming up. Right? Rather than just broadly summarizing what's been spoke about, I want to be a bit more focused. Right? Show me areas of misunderstanding to touch on during my next tutoring session.

So now I have four specific topics that I can bring to that tutoring session, and I can focus around these and make sure that we're we're, moving through some of those misunderstandings, understanding why they're you know, whether it's just a a lack of general experience or if there are differing perspectives that are coming into play, but now I know where I can focus my session to be most impactful. Right? That's one example of, you know, how I can prompt the summarization tool and and use it to to be a bit more impactful in my teaching. There's many others, but that's that's a nice one that I like to use as an example because it really just hammers home that I you know, I could sit down and I could read through a hundred responses, and I could take notes on all of them. But it's gonna take me a while to, you know, personally surface some of the themes and misunderstandings that I might need to tackle in the next session. Right? It's possible, but it would take time.

So this is another area where we're able to give Sarah back time to her day so that she can focus that next session and be as impactful as possible. So, finally, translation. Right? I I can open up the translation here in the discussion forum itself. This is great from a peer perspective, right, peer communication. I like to use that example of, again, if we're thinking about, like, a global online course where maybe there are students from around the world speaking in their own, preferred languages, in the past, would have to, again, copy and paste that, bring it into Google Translate, and then maybe translate your own response back.

And there's going to be multiple layers that are going to dilute the intent of the communication over time. Right? So instead, we can do this in a discussion. I'll look at this in the context of an announcement, actually. And what I can do is I can enable this translation directly within the context of the of the course, in this case of the announcement, and choose the language that I want to translate this to. So I can translate this to Dutch in this case.

So let's imagine that one of my students or maybe a cohort of my students are Dutch speaking and would prefer to communicate in Dutch or are more comfortable, let's say, speaking in Dutch and reading in Dutch, they now have that capabilities at their fingertips to surface that. They now know that all of this specific information about the course running online rather than being on campus, right, without them having to copy this, paste it, go somewhere else, and then and then refer back to it. So this works both from a, both, as I say, in in the announcements, in the inbox, as well as within a discussion forum as well. So when we're thinking more about that collaborative communication, not just the discussion prompt, but the responses as well from my peers. Right? Being able to then go and respond to Omar about, his thoughts about AI, I can respond in English or Dutch, and he'll be able to translate it back to English.

Right? So think about this as that removal of the language barrier or at least weakening of the language barrier without having to to leave Canvas. So that's that's Canvas core. That's the baseline layer. Right? It's all focused on reducing the day to day friction that your your teaching staff run into before a student has even submitted anything. We haven't seen anything about assessment, yet.

We've just been focused on access, assessment building, communication collaboration. Right? We've seen the setup, and we've seen how we can keep that communication manageable. The next step, which is Canvas plus, is a bit further along this journey. So now that the course is running, the question becomes, how do instructors give students the quality of feedback and attention that it deserves at a scale that most institutions are probably operating at right now? So that's that's where we're going to head to next. So same idea as before.

We're going to dive in and anchor ourselves first around the ethos behind Canvas plus. We'll look at specifically the the elements of of Ignite AI that we've woven into the conversation here, and then we'll dive back in with, a slightly different persona this time. Right. So Canvas plus, what is this about? It's about making learning visible, engagements actionable, and high quality teaching easier to sustain. Right? Sustain, I think, is an important piece there as well.

So what comes along with Canvas plus when we're thinking about Ignite AI? First of all, we'll we'll we'll stick on that track of of assessments. Right? And we'll look at the, rubric generator. So the rubric generator is all about drafting effective and consistent rubrics so that your learners, your students have the visibility into how they're going to be assessed. And at the same time, we we're aiming to save educators as much time as possible throughout this process. Right? The next piece is grading assistance built into SpeedGrader.

And this is all about drafting feedback and and scores within SpeedGrader itself. You'll notice that I in both of these pieces here, I've I've emphasized the word draft. And and, additionally, for the question authoring tool, I use the word draft. This is this is not about removing educators from the loop at any point. This is about supporting the workflows and processes that they're already doing today.

Right? And then finally, insights for discussions. Right? So we'll go a little bit deeper on what evaluation looks like and and what the, the alignment of a discussion might look like in terms of engagement, and we'll see that within the insights for discussions. So our next persona here is David. So David is a course leader, right, at a business school. He has two hundred submissions that are due Friday.

Right? He wants every student to get feedback that actually is going to mean something to them, and he does not have much time. Right? It's it's end of day Tuesday. We're getting close to, the end of the day, so he has a couple of days basically to get through this. So let's see how Canvas plus can actually change what's possible for David, how it's going to affect his week, and how that, plays into the experience of his students as well. Right? So where we're gonna start then is with the rubric generator.

We have an assignment here. Right? An assignment that is on lived experiences, systems, and power. We're talking about a sociology course here. Right? And we have the typical information, your learning objectives, an overview of the assignments, instructions to be followed, and so on. Right? But what we don't have yet is a rubric.

Right? I can go and create a rubric. I can find a rubric that might exist already, but it's, you know, possible that that there aren't any rubrics that really fit specifically with what I'm looking to assess here in this case. Right? So what I'll do is come in here and create a brand new rubric. I'm gonna call this just lived experiences. And I'll use this rubric for assignment grading as well once we get to that point.

So just like before, right, I I I equate this to the question authoring tool in a way because I still have all of the the tools that I had before. We've not removed that that approach to rubric creation where I can go and draft new criterion, give it a name, I can get the rating descriptions, so on. I have full control over this. Right? Likewise, I could pull from existing learning outcomes, and I could bring those into the rubric here. So that hasn't gone away.

The only thing that we've done is we have, enhanced what that workflow looks like, and we've added another avenue to the creation of these frameworks for assessments. Right? So within the rubric generator here, I have a number of different configuration tools. So I can decide what grade level I want this rubric to be, you know, within from a from a guide that perspective, how many criteria I want, how many ratings within those criteria, and then the total points. So this this assignment has fifteen points, so I'll leave that as as fifteen. And then I can dive a little bit deeper.

I can give some additional information about the outcomes that I want to be focused on, and then any additional prompt information. Right? So other ways that I can just tweak what I would like the output of this tool to be. Right? I'll leave those blank. I'll just let the tool do its work without me guiding it in any specific direction at this point. So when I go ahead and click generate criteria, what's gonna happen is it's going to take that source configuration and the assignment description.

That's the important part here. This isn't happening in a vacuum. The tool knows the context of the assignment itself so that it can take that and build out a rubric that is hyperspecific to this particular assessment. Right? So I have a criteria criterion on the applications of sociological theories and concepts on sociological inquiry, analysis of systems and power, centering lived experiences, and equitable interventions. So these are my five ratings or five criterion criteria, I should say, that are very specific to this assignment and will give my students a very clear picture of how they will be assessed.

Right? Now just like before, this isn't just generating something and then walking away, generating a a quiz question, generating a rubric. This is directing that process, first of all, reviewing the output, right, being able to come in here and make any changes that I might want to, and making the final call. Right? I'm always going to be in that driver's seat and controlling what is put in front of my students. Right? The real value here is the drafting speed. Right? This type of rubric could have taken a whole afternoon.

Right? Instead, within a few minutes, I've I've drafted a first pass, and I I might still need to make changes. I might need to tweak that to be more focused on, the way that my institution is operating. Right? But that has probably significantly cut down the effort and time that was needed from from my day to get that into place and give my students a clear idea of how they will be marked for this particular assignment. Right. So what does that feed into? That that very, sort of nicely leads into what that that feedback process actually looks like.

Let's take this example, right, of this assignment of lived experiences. Let's take that rubric that was just created, and what let's put that into place in the speed grader. Right? Let's let's fast forward a little bit. Let's assume, that I have a student who has submitted some work. Right? So we have Omar here.

Omar has submitted, an assignment for the, what's this one, an assignment for this particular or a submission for this assignment. Right? I have my rubric down below, applications of sociological concepts. I I ran this before the call today. That's why you you'll actually notice there's already, some of these icons here that are appearing. So if it hasn't been run, then those wouldn't be showing.

But what I can do is I can utilize this auto evaluate button. Right? Now there's always lots of conversation about this particular feature and and do you value that it's really bringing to the table from an educator's perspective. And, I guess, first of all, what is it doing is the first case. What it's doing is it's looking at the assignment itself, right, the submission. It's looking at the rubric, whether that has been generated by the rubric generator or not.

It doesn't matter. It just needs to be a rubric in there to give that structured feedback, and it gives me some draft feedback. Right? So I have information about, which of these levels that are going to be used in here. Right? Some of this will give me the comments as well. Right? So these are comments that have been generated by the grading assistance tool.

Again, this is this is about giving me a a a well structured draft of feedback. Right? Rather than, you know, again, me me going over to another tool and and generating some content there and then bringing it back into Canvas, I can do the all of this within the framework and the focus of Canvas, the LMS itself. And I still always will have that control over what gets put in in front of my students. Right? I have full access to these comments. I can change any of these comments.

Right? Maybe that really doesn't align with what I'm thinking, that particular comment. I can remove that, and I can decide what to move forward with. Right? The output of this tool is the starting point, not the final word. You'll also notice here, I can decide whether or not I want to include comments. So if I if I wanted to keep this more eye level and give me a guide on the rubric, the sort of scoring rather than the AI generated comments, I can do that and then use my own comments to to bring that that more, human connection between myself as the teacher and the student into Planhere.

Right. So we have one submission here. Again, if you remember the scenario I set up for David, he has two hundreds, two hundred submissions he needs to work his way through. So using this to give me a a starting point will definitely save time and hopefully keep him from having to work too too deep into the evenings to get those done. Right? Finally, let's have a look at the the insights for discussions.

So we have a discussion here for the social roots of of climate change, and we have tons of responses. Right? This is a really big course, which makes sense. I have two hundred submissions for that other assignment. I also have lots of engagement in the discussion forum. So, again, I could use the summary like I did before, but this time, I'm gonna go and use the discussion insights.

So the idea here is that we're able to surface patterns across the cohort. Right? So for example, I can come in here, and I can see that, you know, many of these responses were relevant to the discussion topic. Right? That that's what this is all about. It's about connecting the discussion topic to the response that was added. So if I come in here and look at some that are irrelevant, right, I can see that reply here, and I can see that Zara was under the impression that the more than human theory means that climate is actually caused by aliens that are something more than than human.

So that's a that's a misunderstanding in this case. Right? That's probably something that I need to follow on. I can go and see that reply in context. I can get straight to Zara's reply. And I can either reply here or I can, you know, go and and and message, from here as well.

But I was able to surface that of the seventy replies, right, of the seventy replies, four are irrelevant, sixteen need review. So they're kind of in, you know, more of a a gray area, but maybe these are opportunities for me to intervene, open up the conversation, and create, you know, a bit more of a collaborative space to bring everyone up to speed. Right? The value here is really about highlighting and flagging where I can focus my attention to be more timely with my intervention. I'll take a moment. Are there any questions, Laura, that I should Yes.

I had a couple of questions related to the grading assistant, so we'll go back to the previous things you've shown. And some of them, Alex, feel free to say if we need someone else to answer them or if we need to reach out to our product team. But the first one, so when you were showing the grading assistant, you obviously showed the little AI symbol. The question from Jeff is if they were to alter the feedback, would the AI symbol disappear? That is a good question. So if I'm looking at this one here Yeah.

I'm assuming we're talking about the comment here if I, you know, make changes. I believe that that the actual comments, icon will disappear if we make those changes because it's no longer a sort of default comment that's been made by the assist, grading assistant. Also worth noting that I can change the rubric grade here. So where this was something I didn't mention that, you know, this was selected as as two on the grading scale, but I actually think it was a three. The two is a bit harsh.

I always have control over what I'm going to use there for for the grading. So let's let's, follow-up on the icon and where it does and does not appear based on updates that have been made, and we can follow-up with that one. But just a bit more context on sort of Yeah. Your ability to make changes there. Thank you for that, Alex.

Also, the next question is about the grading assistant. Is there any intelligence or learning in the model based on the feedback that a teacher has entered or adjusted, previously? Can it learn feedback styles, format, or language? Yeah. So that's a really good question. It's it's not something that the tool does right now, and that's, you know, predominantly because we aren't training the tools, you know, focused on data that is being entered by you at the institutions. Right? So that that may be something that we explore in the in the future.

But for right now, it's it's not learning based on your own actions. It is, a more static LLM, right, in that way. Thank you, Alex. Yeah. That's true.

And that's also highlighted in the in the nutrition facts, the fact that they're not learning. There's quite a few questions coming up more. Are you happy for me to continue going through them, Alex? So what I'm gonna suggest that we do is we move on to Canvas next and see what that looks like, and then we'll continue back onto the questions at the end just to make sure that we have time to get through everything. Perfect. Let's do that.

Yeah. Okay. Cool. So with that, let's let's jump back into our slides here, and we'll we'll be ending off on Canvas next here. Right? Canvas next is that final tier where we're looking at transforming workflows with AI.

Right? Just from the from the description here, you'll know that this is where we focus a lot of, effort on some of the more, let's say, complex AI tools, not in usage, but more in the ways that they've been developed and the ways that they can actually transform, you know, what what that those experiences look like. Right? So Canvas Next really takes a a a wider view, right, on on interaction and and and education here. So we're moving from the individual instructor in experience, and we're gonna shift more into a view that is that is more institution wide. Right? And then we'll also look at a student's experience. A lot of what we've seen today is more a student inheriting an experience based on what an educator has done.

Right? And some of that might be an educator has more time in their in their day to interact one on one with a student. Right? So there are those implicit, effects that will happen downstream for students, but we will actually see a tool that we've, developed that will support a student themselves in their experiences within Canvas. But for now, let's let's focus on three areas, three functionalities. And these were laid out actually right at the start because they're they're big pieces and important to the way that we approach Canvas next. And that is, first of all, the Ignite AI agent.

Right? So this is something that was launched earlier this year, right at the start of the year in in the US and then a couple of months later, globally. So over here in Europe, the Middle East and Africa. And the Ignite AI agent is all about being able to act on your behalf. Right? That's kind of the key thing here. A lot of what we've done so far is is, you know, surfacing information, allowing the tool to give me a first pass at some things.

The agents means that I can actually harness the power of those APIs, those five hundred plus APIs that that Zach was talking about at the start, and allow the agent to take that off of my plate and do some of that for me. Right? So we'll we'll see how that can, you know, play into, an educator's day, for example. Let's see what that looks like. We have the ask your data tool. This is one that that we launched last summer during InstructureCon, right, within the intelligent insights package.

They ask your data tool, specifically being able to allow you to ask natural language questions and wholesome insight and data from that, right, without having to be a data scientist or a data engineer or have a full pipeline in place to being able to get to that data when you need it. Right? So we'll see that as well. And then finally, Ignite AI study tools. So this is where we shift in the into that student's perspective, and we turn that experience of a student into a more self led learning experience. So being able to create personalized flashcards, practice quizzes, create study aids all directly within the content within Canvas without having to bring that to another tool to let that do, let that do that for me.

Right? So we have two personas to to introduce here. We have Priya, and we have Omar. We we actually saw Omar a little bit earlier because he was, the one that had submitted that assignment that we just graded. But we have Priya, LMS administrator, multi campus institution. She's all about the data.

Right? Understanding what's happening across the institution and getting that information quickly, right, when she actually needs it, not weeks later. And Omar, second year undergraduate student, he's, you know, going through the same things that most of your students probably are today, which is just knowing that they need to keep, you know, working their way through their courses or their modules. Right? They know they need to be revising for, you know, assessment that's coming up, but it's not always that easy to know where to start. Right? So we'll see where Canvas next comes into play for those. Now before we before we jump in with with Priya and her perspective, we're actually going to stick with David for a moment and have a look at the Ignite AI agent.

So the agent is all about orchestrating your own workflows. Right? I like to think about the agents being able to wear many different hats. Right? And I'll I'll go through a couple of those those hats, so to speak, or those use cases that an agent can play, within your day as an educator. So the idea here is that rather than having to navigate multiple screens do things like updating settings or executing executing workflows or anything like that, I can ask for what I want in plain language and have the tool go and do that using those APIs that we spoke about. Right? Which means that action is, secure and transparent, and it's just being done on my behalf.

So I have opened up the agent. I'm sitting on on my course modules here, And I always like to to come in here and just use this option to begin with. So what can you do? How can you assist me with my tasks within Canvas? So the agent is context aware. This is something to be, clear on. The agent knows what it can and cannot do.

That's important. So it can do things like course and content management, assignments, communication, student management. But there's lots of things that it can't do. At least right now, those might be things that we open up in the future. But for now, things like grading submissions, we're keeping that to the, to the grading assistance tool, importing courses, running plagiarism checks.

Right? There's a number of things that the agent for now is is not focused on. Right? From a permission perspective, it's worth noting. This agent has the ability to do what I can do. So it wouldn't be able to go and change my colleague's courses that I don't have access to. Right? It can only do anything that I can do.

So let's go ahead and ask this to do something for us. So the first thing I'm I'm going to do is ask it to create a module. This is gonna be module four. If I condense this down, I have three. But module four is going to be features, justice, and transformation.

I want some core content, a discussion, and an assignment. So the tool is gonna walk me through exactly what it's doing. And the key part here is that before it does it takes any action, it's always going to allow me to approve or reject. Right? So the first action it wants to take is to create the module. So I'm gonna go ahead and approve that.

And then it will work its way through and step by step build out and do exactly what I've just asked it to do. So it's gonna create the module. It will now create a page of content, right, a page of core content here, which I will approve. It will create a discussion and an assignment. So that that checkpoint means that I'm, again, always in the driver's seat.

I can always check and see what's being created, what assumptions might have been made by the tool, by the agent, and I can approve those or reject those. Alright. And then each one of these steps is just being now added to the module in question. So I'm adding the page. I'm adding the discussion, and I'm adding the assignments.

Right? So the if I now refresh this page, oops, I now see that I have module four in there, and I have some core content, and I have a discussion forum that I'm ready to work through. So I I allowed the tool to do that for me. I could have spent a whole day on doing this. Again, this gives me a starting point to work from. This probably is a fairly basic page of content.

Right? Yeah. Very basic. But I could have guided that a little bit more. I was very generic with what I asked it. Right? I just said create some content for me.

I could have been more specific on topics to focus on, how I would like that content to be structured. In fact, that that leads us into our next example here, right, where I can actually use the agent like a an educational consultant. Right? Here's a big discussion. What I wanted to do is actually take this discussion as an input and redesign it so that my students will have meaningful choice in how they're going to be able to respond. This would take some time for me to think through this and decide how I want to reorganize this discussion, how I want to, you know, allow my students to respond in different ways, but I can let the agent do that for me and give me a first pass.

Again, I'm I have control before I go and publish this, but I can allow it to to take a first pass here. So you can see, it's giving me three options. Right? Here's the redesign discussion. So for my students, they'll be able to choose between a a case study, a conceptual deep dive, or systems analysis. Right? I can then just go and say, okay.

This this is perfect. It's now asking me if it would like to me me to update that into Canvas, and then I can say, yeah. Go for it. We're ready to go. That will make that change.

Right? So it's not just organizationally, but also conceptually and sort of from an educational perspective, we can use the tool to be a bit of a sounding board on on how we're approaching some of these things. Right? And then finally, I like to think about using the tool as logistics support sometimes. Right? So I wanna find students who are behind on assignments and create a discussion thread where they can go and ask questions to catch up. Now I could do this manually. Of course, I could go and I could find all the students, and then I could, you know, figure out which ones who's who's behind on what.

I could create a discussion thread. I could sign that that discussion thread. It would it would take time, but I could do it. Instead, I can have this working in the background. I can continue, you know, grading in SpeedGrader if I want to or something like that while I let the agent run-in the background and do exactly this for me.

Right? So it's gonna go and, identify. In this case, there's forty seven students, and that it's gonna go ahead and create that discussion thread for. Right? So these are three examples of use cases. Right? The agent being a course design assistance, maybe an educational consultant, a logistic logistical support person. But the the use cases can really be extrapolated out.

I I like to actually look at this space in in the CAMS community. This is our, our prompt library, and this is all community powered. Right? This isn't, you know, just things that we at Instructure are going and putting in potential prompts in here. And instead, members of our community, institutions that are using Canvas have the ability to come in here and and put in prompts that have worked for them, right, and and share the love around, you know, the ways the agent can can impact the day to day life of of educators and administrators. So this is a really great, resource to look at.

In fact, that discussion prompt redesign, the one that I used, I I took this from the community library. So someone, at an institution that uses Canvas, you know, put that in a month or a month or so ago, and I I gave it a try because I thought it was an interesting perspective. And turns out it works super well, and it's a nice way to, like, use that use that as a way to to enhance and bolster, you know, the types of things that I'm putting out there. Right. So now let's shift into Priya's point of view.

We'll have a quick look at Ask Your Data. Conscious that Ask Your Data has been out for a little while, so it's one that probably people are a bit more familiar with. Right? But let's let's have a quick look anyway just to make sure that we're all on the same page. Again, we're thinking about this in the context of of Canvas Next, right, that that tier of of Canvas. And it's all about using AI to to, transform your workflows.

So the idea here is that what I'm able to do is ask a question with natural language and have it go and query the underlying data and return me an output. Right? I think the key value here really comes from the fact that I didn't have to go and ask my data team who are already overwhelmed with requests from various other teams around the institution to pull data from Canvas for me. All I need to do is ask the question and get a response. As an administrator, this is that has saved me potentially weeks of communication by surfacing that data for me all in one place without me having to pull in people who might not be as familiar with Canvas, for example. Right? I have a full summary of what was what was done, the methodology that was used, any assumptions that were made.

Right? This is really important whenever we're working with AI. We need to understand what assumptions are being made about, the the prompts or queries. And I have the actual output. Right? Here are the courses. I was looking for the top ten courses in terms of activity.

I've specifically defined activity as discussion replies and and submissions, and here's my sorted list of those top ten. I I could go and look at the SQL if I really wanted to. You don't have to. That's kind of the beauty is you don't need to know SQL to get this type of output, but it's there and visible for you if you want it. And I could ask another question, something like how many actively enrolled students are participating in discussion boards each week? I wanna get an understanding of how how collaborative and and communicative are my students across the institution week by week.

So this will not only, of course, return me that data, but it'll show you how we can take that data and and visualize it directly within Canvas as well. That's something that, people have have asked for for a very long time. I started in twenty nineteen, and that was already a question that was being asked is how do we visualize our data directly within Canvas? I I don't have the resources to get this into a data warehouse and connect Power BI or or Tableau. So I can come and create a line chart that will be by week looking at the active and participating students, and I can see effectively a a timeline of what that discussion interaction looks like over time. Right? Finally, of course, we can then take those.

We can pin those. We can create our own, you know, personal dashboards and share those across, you know, our various colleagues. But the key driver here really is around that that natural language prompt and being able to keep everything in one place. Right. Finally, let's let's turn our, attention to Omar.

Omar has been waiting patiently here, for us to see what he's he's up to right now. And Omar is gonna take a crack at revising for an assessment in his climate change and society course. So I'll come into my modules, and I'll go to the core content within module one. Right? So I have content on sociological frameworks for climate emergency. Now, traditionally, Omar probably took this and put it into ChatGPT and asked it to summarize and give him an idea of, you know, what he should focus on, maybe how to create some questions for him.

But that that means he's having to leave Canvas. Right? He's going to some other tool that, doesn't have the oversight of the institution in terms of transparency. Right? Study tools is what will fill that gap though. Right? This means that we can make those experiences available and secure and transparent where the content is already living. So I can summarize this content.

I can create flashcards I can start, just doing some ad hoc revision on this content. Right? So what are the key, sociological frameworks? Right? Have those. We can move move through my my flashcards. And, of course, I can quiz myself as well just to make sure I'm actually understanding what's going on here. I don't need my teacher to go and create a a an actual quiz within Canvas.

I can just have the tool create some questions for me to keep my my revision moving and and making sure that I, myself, actually understand what's going on. Right? So to to wrap everything up here, let's bring it right back to where we started today, where, you know, you've seen a lot, hopefully, the past forty five minutes or so. Right? Much of this already exists within your Canvas environments today. Right? And the key thing is is none of it's a major change to how your staff are teaching or how your institution operates. Right? These aren't massive diversions from your workflows today.

It's about supporting and sitting alongside your existing workflows, you know, not not replacing those workflows. So what the idea today was is to just support that understanding of where the value comes from throughout the different tiers and the different experiences that that leverage AI. And, hopefully, you can come away with this identifying, you know, one or two of these, right, pieces that we've seen, where you might wanna have a closer look and and maybe switch it on in in a space within your Canvas environment, do some testing. What we've seen works really well is that, you know, starting out small, starting to, you know, build your way up with a couple of use cases, that's a good place good place to start. Right? So I know that we have a few questions left, that have come up.

I'm I'm sure there will be a few more that probably come up as well. Let me just pop back over to the to the slides here, and I'll hand it back over to Laura. Thank you, Alex. Yes. So there's still a couple of questions in the q and a, so feel free to add more.

We'll get whatever minutes we have left at the end, we'll use to answer as many questions as we can. And if we don't know the answers, we'll let you know as well. Just so you know, what we're gonna do with the questions that we have not been able to answer, there's a couple of them that are more related to further developments or specific details about the product that we need some further answers to. We are going to be collecting all of the questions from the q and a. We're gonna be creating a q and a as well, a q and a document that we will be sharing alongside the recording for everyone who registered.

So the questions that we can't answer now, we will find answers to and make sure that you all have access to those answers after the session finishes and we share our resources. There's a couple of things we wanted to point out. Actually, earlier this week, we launched a new blog series on the community all related to purposeful AI. So if you actually, I have a link. I'll drop it in the chat later.

But you can read all about what this this particular series is going to be about and when we're going to be publishing different articles for the rest of the month. Every Monday, there'll be a new article published. You can follow the artificial intelligence in education blog in order to receive the newly published blog posts. And the most exciting thing is once we actually finish this series of blog posts, we're also going to publish a free course, that we're going to add to the training portal on Canvas. If you don't know what the training portal is, feel free to ask your Canvas admins if you're not the Canvas admin yourself, because that should be available on Canvas.

And if it's not, that might be a central decision by your institution not to make it available. So we you can address that with your CSM as well. Also, shameless advertisement here for CanvasCon, which is happening in November. If you haven't signed up yet, please follow the QR code. You can register there, and you can still share a proposal if you have an idea for a session that you'd like to present yourself.

I wanted to clarify one thing because we've talked a lot about tiers. And just to ease all of your minds, if you do not want to make any changes to your contracts, you really don't have to. We're not gonna force anyone to migrate. You don't have to move to any different tiers. What you currently have will simply be renamed to Canvas core, and you will continue having access to that.

You will not have to make any contractual changes, and we will obviously continue to develop Canvas core in the same way that we have in the past. So I just wanted to clarify that for some confusion that might have might have come up. I'll go back to the q and a now, and I'll run through the questions that are still outstanding. If you need to drop off, feel free to do so. But if you can stick around a couple more minutes, then, maybe some of the questions that you've raised will be addressed by us.

So thank you so much if you're gonna drop off for joining us today. Really appreciate your time. It's been a long session, and, obviously, we hope to hear from you soon. If you have any feedback around the session, feel free to reach out to your CSM as well and raise your feedback there. We want to make sure that we improve and that you have all the information that you need in order to confidently use the tools that we're offering.

Well, having said all of that, let's move to the outstanding questions in the q and a. So the first question that we've got on here is actually a question I wanted to answer live. I don't have the answer right now, so it's going to be one that we're gonna add to the q and a afterwards, which which is whether or what the progress is of the block editor and whether that will be available in Canvas core or if that will be associated with Canvas plus or Next. I want to make sure we give you the factual correct answer, so we're gonna answer that, later in the q and a that we'll share via email. Then the next question we've got, will the institutions be able to add context on Ignite AIs? Things like writing style, institutional rules, grading rules, etcetera.

So I think maybe, Alex, you can respond to a part of that question. I imagine we may need to involve product as well to give further context. Yes. There are some areas where we can add institutional level context. Ask your data is one of them.

So you can actually, there is an area of the ask your data tool where you can go and kind of create, like, a context library that's specific to your institution, like policy documents or a student handbook. Anything like that can go in and provide that context, that's specific to ask your data tool. The individual tools have ways to do this at the course level, right, or the usage level, let's say. So we saw where we can tweak the prompts for the rubric generator and the question authoring tool, But we're still looking at what that looks like in at an institutional level, so that is likely to be something at a further phase down the line where we start adding institutional layers of control and context over some of these things. Yeah.

Thank you, Alex. So the next question, which I think is an interesting one, have you seen AI hallucinations affecting the Ignite AI agents? What sort of things should I expect to see? And before, Alex, maybe you've tested with it, you share some of your personal experience, this is a good prompt for me to clarify that at this point in time, if your institution allows it, you are able to test out the Ignite agent for free, and you will be able to keep doing that up until September thirtieth. So feel free to switch it on, maybe on a subaccount, maybe on a course, and you can test it out for yourself and figure out what kind of potential hallucinations you might see or things to be aware of before you launch it out to your end users. But just in case you don't wanna do that, obviously, maybe Alex can talk a little bit about his personal experience. In terms of testing, what I just said goes for all Ignite AI features except Intelligent Insights Ask Your Data.

That's currently already a paid add on, so you won't be able to test that one for free. The student study tools, which we didn't really look into deeply today, are also not something you can test out. So that will only be launched as part of the page Canvas next year later this summer. So just wanted to clarify that. But, yes, Alex, I don't know if you wanna share anything from your personal experience.

Yeah. I mean, anecdotally, I've used the tool a lot, in various demos and just my own my own personal exploration. I've not experienced any hallucinations. I would imagine the the bulk of that is because of the way that we've given the model the context of the APIs and and have sort of strict guidance on on how the tool is meant to operate rather than being a sort of wholesale LLM, like something like Claude's. Right? We focus that use case, so it makes sense that that that quite significantly cuts down on hallucinations.

So, yeah, anecdotally, I haven't experienced any. But as Laura said, it's it's always a good place to start in terms of testing. Play around with it. Give it a go and and, you know, see how it how it interacts. But, again, we always have that approval checkpoint, which is always in place.

So that's that's the sort of final final gateway that you can use as a safeguard. Thank you, Alex. I know we are out of time. So the remaining questions, there's still a couple outstanding. We'll make sure to address in the general q and a document that we share afterwards.

So thank you so much all for joining in your time, and we look forward to chatting more. Thanks, everyone. Bye bye.