Episode Transcript
[00:00:00] Speaker A: This week I'm really excited to have Michael Yaffe, the co founder of a company I've written about a lot, who we've worked with by the name of aist. And so, because Michael and his team are so innovative and so creative in the L and D space, I really wanted to give him a chance to talk and explain what he's doing. Michael, why don't you take a couple of minutes and just tell us the story of AIST and how you guys came came up with all this and then you can tell us what it is. Of course.
[00:00:29] Speaker B: Yeah, totally. Yeah. Josh, first up, thanks so much for having me. Yeah. So zooming out a lot. ERIS started a few years ago initially as a way to deliver learning in war zones. So we were kind of obsessed with this question of how do you deliver learning and how do you actually help make people better in really hard to reach locations. So we started delivering learning via text message initially just to make learning more accessible. And we started realizing that by delivering learning by text message and meeting people where they are, we are actually creating dramatically better learning outcomes as well. Right. I think one of the biggest challenges with learning today is that on average, you need to go through about 17 clicks to get to the right content in the right content system. Right. So if you are a sales rep at an organization and you want to rapidly upskill yourself on a piece of content or on a product, unfortunately a lot of learning platforms today make it incredibly difficult for you to do that.
[00:01:18] Speaker C: Right.
[00:01:18] Speaker B: And I think the bigger issue is that a lot of people don't know what they don't know.
[00:01:22] Speaker C: Right.
[00:01:22] Speaker B: So the expectation today is, hey, the LND team is going to create a bunch of content and then people are going to go and find it.
[00:01:28] Speaker C: Right?
[00:01:28] Speaker B: But the reality is that a lot of people don't actually know what they don't know, especially in the critical moments of need that they have.
[00:01:33] Speaker C: Right.
[00:01:34] Speaker B: So we started realizing that by meeting people where they are in delivering learning by tools like text message or Microsoft Teams or Slack, and delivering learning as a chat, as a conversation with a friend, instead of as a video or a long form elearning, a few things happened. One is learning became a lot more bite sized, a lot more digestible and spaced out over time. If anyone's ever used a duolingo, it's the same research that makes a duolingo really successful, but applied in a corporate setting.
[00:02:00] Speaker C: Right.
[00:02:00] Speaker B: So, you know, fundamentally outcomes like behavior change, knowledge retention got a lot better. The second thing that got a lot better is that we could now actually push the right learning to the right people instead of having people go somewhere else to learn. Right. The reality is that today organizations have a ton of data on what somebody needs to know, right? I know that if you are newly promoted as a manager, you probably have to, in month three of your role, have a difficult performance conversation with one of your team members. But today we expect managers to go find that information when instead we could just push the right information to the right person at the right place in the right time. So you can think of Aeris essentially as a really easy way to very rapidly create, using AI, the right courses and the right nudges for the right people at the right place at the right time, automated personalization and localization and translation. And then we use internal data from the H R S, from the lms, et cetera from the CRM to push the right learning to the right people at the right place at the right time, directly in the messaging tools they use every single day. You know, this has now been adopted by dozens of Fortune 500 companies. And the organizations that use Aeris on average see about a 90% satisfaction rate for eLearning by ARIS compared to like 10% for every other medium. They see about a 5x increase in overall learning engagement, even for fully optional courses. And the thing that we're most proud of is there's now been research from Stanford and the University of Washington that proves that ARIS actually meaningfully increases performance for everybody from sales teams to managers.
[00:03:30] Speaker C: Right?
[00:03:31] Speaker B: So if you are an L and D organization that really deeply cares about actual actually improving performance for your organization and improving the performance and outcomes of the people that you work with, you need to change the delivery mechanism and you need to meet people where they are. And that's really where we come in.
[00:03:46] Speaker A: Okay, so learning in the war zone is basically where we all are. We're all basically in a war zone for time because we're getting distracted by 10,000 things. So our experience with ARIST is exactly what you said the Net promoter of the courses we built on aorist are much or not much, but they are higher than the traditional cohort based training we do in the academy. But the pushback or the question that I think people have is you might consider this to be micro learning, which I know you guys may not like that term, but I want to give you a chance to talk about it and you might consider it to be complementary or supplementary to something else, which is considered to be the fundamental learning. You have argued with me multiple Times that that is not true. So how do you deal with that issue of is this core and deep enough to really train in somebody in something that's really critical?
[00:04:38] Speaker B: Yeah, excellent question. So, so a few things that I'll say to start, right? One is ERIS is not going to be an effective way to teach you how to fly a plane. Right? There are fundamentally use cases that are very, very complex, that need to be very hands on, that need to be in person. Our job is not to solve for those use cases. The way that we think about it is that if you zoom out and look at the landscape of, you know, the content that an enterprise learning team delivers, 20% of those use cases are super hands on and super in depth. But the reality is 80% of the learning that an organization delivers is really foundational. It needs to be delivered quickly and doesn't need to be an hour long elearning. Right. Like I'll use a sales team as an example. If you are training a salesperson, 20% of their training needs to be very, very hands on product training so they understand how to use the product and you know, very hands on methodology training, so they deeply understand the methodology. But the reality is that on an ongoing basis, you need to very, very rapidly send people competitive updates, product updates. You need to constantly refresh the methodology. You also need to, from 0 to 1, train folks on critical sales skills that they may not have or that their manager has identified. And so when you're delivering learning at scale, for a lot of updates, a lot of core foundational trainings that are really simple in nature, and for a lot of foundational upskilling, AERIS ends up being a much, much better and more efficient way to do it. Right? The way that I think about it for the, for example in the context of the Academy, is that if you are an HR professional, you need to go through the really, really in depth courses that JBA offers. Those are critical, they're incredibly well done, incredibly impactful. But if you need to very, very rapidly, in the exact moment of need, upskill yourself on how agents are going to affect your role as an HR professional, that needs to happen immediately and you need to be able to upskill yourself in the flow of work very fast. And that's those are sort of the use cases where ERIS becomes very impactful.
[00:06:27] Speaker A: And the fact that sometimes you don't have a lot of time, so you might have 10 minutes here or five minutes there. So given the fact that the AERIS content is so dynamic, I want to ask you a little bit more about sort of the process of building it. But how does an L and D professional keep it up to date? And I know the answer to this, but I want you to explain it because you're talking about a more. A much more dynamic experience that could be changing very rapidly or regularly.
[00:06:51] Speaker B: Yeah, totally. So in Aeris, you can update content very quickly, either by asking AI to update it for you. So we have an AI tool where you can add in feedback, add in updated content, and our AI will actually update the course for you within three minutes. Right. So that's the really fast way of doing it. We also have an authoring tool built in where you can manually go ahead and adjust or edit content. Editing a course in Aeris is as easy as editing a LinkedIn post. So our goal is to make the speed of creation very, very fast and the speed of editing very fast as well.
[00:07:18] Speaker A: So if I. So if I have a bunch of salespeople and they've been trained on something and then there's an update, if I push the update out, will everybody see it immediately?
[00:07:26] Speaker B: Exactly. Yeah. Everybody will see it that exact second.
[00:07:29] Speaker A: You know, we did. We. We've been doing a lot of research on the use of AI, and one of the things people don't understand is I think the biggest use case or benefit of AI is actually speed. The quality is probably equivalent or maybe slightly less, or maybe equivalent or more, but it's the speed at which you can get things into the market to update content and make it relevant and more interesting. I think that's more, more valuable than anything else. And that's kind of where you guys came from.
[00:07:54] Speaker B: I completely agree. I think so. Today, on average, it takes a typical L and D organization 8 to 12 weeks to get the right content out. Right, Right. So. So let's say, keep using the sales example, right? Let's say you have a sales leader that has shared a meaningful update with the team and you need to upskill all of your reps on this new market update. For example, or like a regular regulatory change today, it's going to take an L and D team 8 to 12 weeks to build that content, get it through a review process and push it to the right people or deliver it live. We had a client, a large pharmace organization that start to finish, that they had a Medicare Part D training that they needed to get out to people, start to finish, they were able to upload all the relevant Medicare Part d content. So 600 pages of documentation, they're able to upload that, have our AI build a course, have Rai automatically reference it, automatically translate it into 30 languages and push it to the right people start to finish in under 30 minutes. So everything from need to review.
[00:08:52] Speaker A: I also want you to talk about what you did with Sidekick or the tool that. That can actually do upfront needs analysis. Explain that, because that to me is really interesting.
[00:09:02] Speaker B: Yeah. Yes. So that to us is sort of the next frontier of where AI can really make really save a lot of time.
[00:09:08] Speaker C: Right.
[00:09:08] Speaker B: So if you look at how LLND teams spend their time, most time is spent on delivery creation and needs analysis. So we've automated delivery with delivering learning, you know, via text message and teams chat and Slack. We've now automated creation with our AI creation tool. The next step is automating needs analysis. And so we've developed a tool called Teammate that will actually go out and via voice chat, interview stakeholders internally. So you can say, hey, we're launching this new product. Can you go interview these five stakeholders on our product team? Also can you please interview these three stakeholders from our sales management? And then, you know, it'll. Our tool will go out, interview people, aggregate all of that data, and then use that to automatically build a course. So all you have to do as an L and D professional is tell Teammate who to chat with.
[00:09:57] Speaker A: For people listening to this podcast, I want to really reinforce what Michael just said.
And by the way, this applies to recruiting and org design and change management and a whole bunch of other things HR people do. So in the needs analysis phase, where you're interviewing people, visiting people, asking managers questions, the Teammate tool can, I guess, you have to teach it a little bit about what questions to ask. Right. And then sort of. And you give it a list of people you want to interview, and it literally goes out and sends them messages to get the data back through. Through voice or text, I assume. Right?
[00:10:31] Speaker B: Yeah.
[00:10:31] Speaker A: Consolidates it, shows you what it learned, builds the instructional plan, allows you to iterate on it, and boom, your needs analysis is done. Correct? Yeah, I just think that is such a huge idea. And you guys are the only ones that I've seen that have done that, that upfront part.
[00:10:49] Speaker B: Yeah. Okay.
[00:10:50] Speaker A: So, yeah, so I'm sure you guys are growing like crazy and you're getting a lot of customers. What about the lms, the backend, the infrastructure? How does all that work? Because this is such a new paradigm.
[00:11:01] Speaker B: Yeah, totally. So for us, currently, what we're doing is we're aggregating all the data in eris, we have a ton of data Analysis built into the platform and then we can send that data back to your LMS so that everything is in one place. Right. So our goal is that for a learner, most learners don't know that Aeris exists. Right. They're just getting, you know, messages directly in teams, chat or Slack, and then all of the data is automatically synced to their learner profile, you know, in the LMS that the organization uses. So most of our clients sort of layer ERIS on top of their LMS currently.
[00:11:28] Speaker A: You know, we were just talking before we got started recording that the general L and D world is a little bit behind on this new stuff. I think AI has been very widely adopted in recruiting and internal mobility and skills. And unfortunately L and D people haven't had these kinds of tools in the past. And you were making a comment a little bit earlier that a lot of your business is coming from line leaders and the L and D people are either blocking it or slowing it down or just taking too long to. I have my own opinions, but what do you think is going? How do we get L and D to understand the value of this stuff?
[00:12:00] Speaker B: Yeah, totally. So unfortunately, some of the L and D teams that we chat with, you know, there's a lot of very, very successful, very high performing L and D teams to work with. Some of the L and D teams that we chat with are very, very tied to SCORM files as a delivery mechanism or virtual sessions as a delivery mechanism. Right. And so what happens is that when you're tied to those mediums as a delivery mechanism, you're tied to this like a ton of technology that doesn't actually do what you want it to do.
[00:12:24] Speaker C: Right.
[00:12:24] Speaker A: And from my point, that's because they're afraid of losing their jobs potentially.
[00:12:29] Speaker B: I think that may be a part of it. I think it also may be the fact that change managing within large organizations is difficult.
[00:12:35] Speaker C: Right.
[00:12:35] Speaker B: Like saying that, hey, you essentially have to acknowledge that, hey, the way that we've been deliver learning for a very long period of time actually isn't delivering the results that we want. And that's a very hard conclusion to reach. And it's a very hard conclusion to say out loud. But I think that's kind of the conversation that needs to be had.
[00:12:50] Speaker A: Yeah. And you know, and just for the people listening, I mean, that's exactly what happened. When Michael and I first met a year or two ago and we started using it, we had our own people inside our company that were skeptical, but we immediately saw the value and you know, it Picked up speed very, very quickly. Are you guys the only ones doing this? Who else is doing this? I mean exonified, did they do that? Who do you see out there? That's this in the flow of work, micro learning, real time updating stuff.
[00:13:15] Speaker B: There's a, I think there's a handful of companies that deliver learning via an app, but we just have found with many of our clients that app based learning doesn't have nearly the same impact as directly meeting people where they are. Right. So yeah, we're the only company that delivers learning fully in messaging tools. And by fully I mean like the learning happens as a chat conversation.
[00:13:34] Speaker C: Right.
[00:13:35] Speaker B: And for us, our vision is really to build this very, very seamless and integrated way to get from business need to business outcome as fast as possible.
[00:13:42] Speaker A: Right.
[00:13:43] Speaker B: And so we' really the only ones that are building that end to end experience.
[00:13:46] Speaker A: You know, what do you think's next in terms of AI stuff? You guys are, I think you're at kind of the cutting edge as it is. But where do you think this is going to go if you can share a little bit about what you guys are working on next?
[00:13:55] Speaker B: Yeah, the way that we think about it is that there will be AI agents for L and D teams. Right. That help L and D teams. Our idea is that if you are an L and D leader, you should be able to wave a magic wand and an AI agent should be able to execute everything from needs analysis to delivery to data analytics for you. Right. So teammate is sort of what accomplishes that on the lng.
[00:14:16] Speaker A: Michael, let me interrupt for you one second.
[00:14:17] Speaker B: Oh, totally. Yeah.
[00:14:18] Speaker A: What about the scenario where a user is consuming content, taking a course and they ask a question, can't you just open up a chat and ask a question of the corpus? So this thing becomes an intelligent agent really, not just a training delivery system.
[00:14:34] Speaker B: Exactly, yeah. So the goal is that within, you know, the long term vision for us is that not only do you have an intelligent agent helping the L and D team in the business unit achieve their business outcomes as fast as possible, but you also have an intelligent agent working with the, with the learner constantly sort of reviewing their data and helping proactively pushing learning to them. And then when the learner reaches out for, for, for help, then provides sort of feedback and guidance. I think the really critical piece here is that a lot of people assume that chatbots are sort of the end all, be all solution for, for like knowledge and for learning. What we've noticed is that most people don't actually know what to ask in AI. Right.
And so a major challenge a lot of organizations have is they'll deploy a tool like ChatGPT internally, and then they see very, very low utilization adoption rates because employees don't know what they need to learn.
[00:15:24] Speaker C: Right.
[00:15:24] Speaker B: And so one of our core hypotheses and one of our core belief beliefs is that we really believe in pushing the right learning to the right people. And so the goal is that continually, eventually sort of add in more support for people being able to poll learning and ask questions. But the core question for us is going to be how do we consistently push the right learning to the right individual based on all of the data we have about what will it take to make this rep successful and how can we give them the right information to do that?
[00:15:49] Speaker A: I think that's absolutely spectacular. You know, we've seen exactly the same thing with Galileo, and so we've been training Galileo to push responses and inquiries back. It's almost as if if you asked a chat system a question about a problem or you were learning something and then you went back to your job and didn't use it for a few minutes, maybe the next day the system would, hey, yesterday you asked me about such and such. We have some new information on that. Would you like an update?
[00:16:12] Speaker B: I love it.
[00:16:13] Speaker A: Yeah, I think that's where this has to go. Okay, Michael, is there any, you know, I don't want to spend too much time because we have only 20 minutes, but what do you think the biggest learning you've had for people new to this to get started? And it may be as simple as just give it a try, but where should they start? What is the right application area? That seems to be a home run for day one.
[00:16:35] Speaker B: So a few quick notes. One is, from our point of view, the right application on day one is usually something business critical. Right? So something like AI upskilling or sales training. We've seen a ton of success with product training. In particular, we highly recommend, like something very tactical, very tangible is where we recommend starting to. The second note is put all of your preconceptions about AI aside and actually see what the tool does, because there's a lot of misinformation about, about sort of the state of AI. One example is that if you look on LinkedIn, a lot of people are saying, hey, AI is still hallucinating. You can't trust AI content. That' no longer the case. AI is now actually more accurate than most human learning designers and with referencing can actually cite every single claim. It's making.
[00:17:14] Speaker C: Right?
[00:17:14] Speaker B: So put all of your concerns aside. And three, a lot of organizations, a lot of L and D organizations in particular, are scared of their IT team because they have an assumption that their IT team will block a lot of AI. I can only speak for Aeris, but I know that internally we've had no issues with IT teams approving Aeris because most organizations like Aeris architected the AI in a way where it's very, very data safe.
[00:17:35] Speaker C: Right.
[00:17:35] Speaker B: And where it's designed to meet sort of enterprise safety. So move fast, find a compelling use case, and put all of your preconceptions aside. This is really the moment for L and D from our point of view.
[00:17:46] Speaker A: Well, I couldn't agree more. I mean, we've had just spectacular success with you guys, and I think this is just. You're really ahead of the curve here. So. Michael, thank you. Okay. It's a R I S T. The website is a R I S T CO thank you again, Michael, and we'll talk again soon.
[00:18:03] Speaker B: Awesome. Thanks for having me.
