Name: 2026 Trends: Solving the AI Productivity Paradox
Uploaded: 2026-01-14T19:16:06.986Z
Duration: 1 h 8 s
Description: 2026 Trends: Solving the AI Productivity Paradox

Transcript for "2026 Trends: Solving the AI Productivity Paradox": Hello, Good Morning, Good Afternoon and Good Evening and thank you all for joining us today. It is great to see so many people here from all around the world. I am Shelly Kramer, Principal Analyst at Kramer and Co and I am thrilled to be moderating this session today for a dive into Grammarly's recently published 2026 AI Shortlist Report. The report explores three critical trends that separate AI activity from AI impact, which we call the AI productivity paradox, and the shifts that are happening as a result. Our discussion today features a panel of two founders who will explore the importance of embracing an AI native mindset and the opportunities for reaping the AI related productivity benefits that exist for organizations who do. We have got a full lineup today and a fascinating discussion ahead. It is going to be a fast paced discussion with multiple perspectives, so we will move quickly and we will cover a lot of ground. Before we get started though, let me cover a few bits of housekeeping. Because of the size of the webinar, all attendees are automatically muted. To enable closed captioning, click the CC button at the bottom of your webinar window, which is pictured here. We would love to hear from you throughout the session, so please leave your comments in the Q and A widget. We have team working behind the scenes to help answer your questions. And don't worry about taking notes, we will follow-up with a recording of the event so you won't miss a thing. So here is what we have planned for today's session. It is going to be a full and fast paced hour. We will start with quick introductions to our panel of founders who each bring a unique perspective on AI native productivity and the future of work. We will touch briefly on the evolution that Grammarly has undergone over the course of the last year or so and what is ahead on that front. Then I will set the stage with some context from Grammarly's 2026 AI Shortlist Report specifically what we mean when we talk about the AI Productivity Paradox and why it matters so much right now. From there, we will dive into our panel discussion about the trends laying the foundation for AI native work. I want to thank everyone who submitted questions in advance. We are going to try to cover as many of them as possible. And finally, we will try to leave room for Q and A, so please drop your questions into that Q and A widget throughout the session. Now, let's get to the stars of the show. It is my pleasure to be joined by Shashir Ramota and Rahul Fora and I am sorry to report that Max Litvin, one of Grammarly's co founders, had originally planned on joining us today, but he is not feeling well, so we are going to soldier on without him. Let's start with brief introductions from each of you. Shashir, why don't you go first? Sure. Thanks so much, Shelley. I'm, Shashir Mrotra. I'm the CEO here at, Superhuman, formerly known as Grammarly. Before joining the company, I was the founder of Coda, makes an all in one, productivity suite that was acquired by Grammarly. Before that, I spent a number of years at Google, most of that time running the YouTube group, and before that, I got my first crack at productivity working at Microsoft. Awesome. Awesome. Rahul, what about you? Hello, everybody. I'm Rahul Vora. I was the founder of Superhuman back in the day. We were acquired by Grammarly, now collectively, along with Coda, we are known as Superhuman. Today, I lead our Superhuman mail business where we make the most productive email experience in the world. Most of our customers get through their inbox twice as fast as before. They respond faster to the things that matter, and they say four hours or more every single week. I love it. I love it! Thank you gentlemen. You both bring a wide range of different skills and experiences to this conversation which should resonate with our audience in myriad ways. Before we dive into the meat of our webinar though, there have been some big changes at Grammarly over the course of the past year, so acquisitions and name change and some really exciting things ahead. Shishir, will you walk us through the highlights and some of your thoughts on what's ahead for the company? Yeah, I think it's felt like a lot of change here, I'm sure for all of you it's also seemed like a lot of change from the outside. So I can give a little bit of a history of what's happened in the last year. So as I mentioned, I was the founder of a company called Coda, We were acquired into Grammarly and I became the CEO of Grammarly at the time. And then I decided to continue that by acquiring another company, the company formerly known as Superhuman, that's the company that Rahul founded. And we started a new product as well, a product we call Go, which I'm sure we'll talk more about. Probably one of the biggest things that happened is we decided to change the name of the corporate entity. We were looking at, we had four products in the portfolio going from one and of a sudden, and the corporate entity was still called Grammarly. But Grammarly actually refers to a very popular important product in the portfolio, but probably the least stretchy brand of the set of them. And so we went looking for a good brand for the company that houses these four products. And we decided that the best brand we could find was a name we really, really love called Superhuman. So now the whole company is called Superhuman. We have four products underneath that, we call them Coda, Mail, Go and Grammarly. And all of those together become our AI native productivity suite. I love it. Awesome. Anything to add to that, Rumul? Sure. I would echo everything Shasher just said. I would also add that I've personally never seen a company go through so much change so quickly. Like you said, just over a year ago, a year isn't a long time, we had a different name, we had not acquired Superhuman, we had not acquired Coda, we had a different CEO, we were a single product company, which is maybe a bigger change than it sounds. And during that time, we've made such tremendous progress, not just on the corporate structure and what we're called, but actually all four of our products themselves also. To give an example from Superhuman Mail, when we first launched what we call Superhuman AI, it was just a simple prompt that would attempt to write emails for you. Today, it is an AI agent that can use multiple tools, resolve ambiguities, and think for as long as it needs to write compelling emails that you would actually send. And the proof is in the pudding. Folks are sending these emails. In 2024, for example, we doubled the percentage of emails sent with AI. In 2025, we doubled again. And in 2026, we aim to double at least once again. And that's just one of the four products that we now are as a company. In each of those four products, we've made very similar, if not larger amounts of progress. It's just such an exciting time to be here. You know, it is a really exciting time. Thank you both for your brief overviews and I will tell you that from the outside looking in, I have been a customer of Grammarly's, I have worked with the team at Grammarly for a number of years, I have watched all this change and these evolutions take place, acquisitions and everything else, and the thing that sticks out to me is that as a company from a cultural standpoint, you've managed to hold onto your humanity and, you know, the company has grown, so much has changed and everything else, but there's still the warmth and the passion for customer service and the passion for excellence that's always been a part of this company from the beginning. So I think that's really important as well. Congratulations to you all! Thank you. One comment on that, the name Superhuman, as we looked at different names, one of the reasons we really fell in love with the name Superhuman is, first off, it's a much broader name, it covers breadth of the things we're doing, gives us license to build out the full suite, we think. But actually it's the second half of the word that we really fell in love with, it's the human part of what we do. And if you think about a world full of people working furiously on taking advantage of this AI wave, bringing it to people in lots of different ways, many of them are focused on replacing the human. We're the company that looks at it the opposite way. We look at it as our job is to empower the human. And if you just think about the way Grammarly works, Grammarly has long been the tool that brings the best out of you. At the end of the day, you still, we give you lots of assistance, but you actually send the blog post or you send the email or you submit the essay. Our job is to help you be a better you. And so from that perspective, one line I really like is to think about it as Grammarly spent the last sixteen years turning people into super writers, and now we get to spend the next few decades turning people into super humans. It's one I love way to think about what we're doing. I love it. I think that's awesome. Well, it's no doubt been a very exciting year for Grammarly, now superhuman, and I have no doubt there are good things ahead for the company, for your customers and beyond. So with that, let's get started. So before we get into the trends, I want to ground us in the problem that sparked this report, the AI Productivity Paradox. This report hits on the chasm between AI activity and AI productivity that is a reality across the workplace today. Results on productivity are asymmetrical and they are wildly unpredictable and in some instances workers are actually finding themselves more productive, but in other instances we are finding people are less productive as a result of AI. This is not, however, an AI problem. It is a context problem, an integration problem and a workflow problem. Treating AI as an AI add on or an afterthought is very common, but that is also highly problematic. What we are seeing across the business landscape is that successful organizations are embedding AI into the flow of work and rethinking workflows with AI built in as a collaborator from day one. We are going to remove the slides now so that you will be able to see the handsome faces of our panelists. Gentlemen, are you ready to dive in? Sure. Alright, let's go. Alright, the AI productivity paradox is something most leaders are feeling. Rahul, I know this is something you are deeply immersed in. Why do you think the promise and the reality still feels so misaligned? Well, the paradox is this. We all know that AI is changing everything and rapidly. We all know that we're meant to be investing in AI and rapidly. And it is very likely the case that your competitors are doing this right now. And if they're doing it well, they're probably starting to win. But like you said, according to an MIT study from last year, 95% of organizations are paradoxically seeing zero returns from their AI pilots. So what's going on here? How can companies be winning but 95% of organizations see zero returns? Well, you might think the problem is that people aren't using the AI. Maybe it's a usage problem, but that's actually not it either. In fact, another report from McKinsey found that employees use AI nearly three times more than managers think they do. The problem is actually this. Think about your day to day usage of AI. We, first of all, have to remember to use it. Then we have to stop what we're doing and switch to a chatbot. Then we have to figure out how to craft a good prompt. Then we have to iterate a whole bunch. And then finally, copy and paste those results back to where we started. We call this the AI detour, and it obviously massively limits impact. Instead, what we need are what we call AI native tools. And in particular, AI native tools that have three pillars. They are number one, ubiquitous, number two, proactive, and number three, connected. By ubiquitous, we mean that they work in all the apps and the websites you already use, just like Grammarly does today. By proactive, we mean that they help without you having to ask, And by connected, we mean that these tools know what you know and can actually take actions on your behalf. That's when the AI starts to feel less like a tool and more like a teammate and a collaborator that you can work with. Yeah, absolutely. Well, you hit the nail on the head and that is the paradox in a nutshell, and it's what led us to those three trends in the 2026 AI Shortlist. The report highlights three shifts that we believe separate activity from AI impact: Context, Integration, and Ubiquiti and those AI native workflows. What is important is that these are not new or abstract ideas. Each one reflects how the founders, how you originally built your companies and that is what has become increasingly clear in the past year or two is that real productivity gains emerge only when these principles come together. And that convergence is what is driving the next chapter of AI at Work. That's why I decided to get into these trends, just to understand the ideas themselves, but to hear how they started and where you are taking them next as one company with Superhuman. Let's start first with Context, the first trend in the report. Let's talk about why context with AI is important and what it actually means. Grammarly was built around the idea that AI only works when it understands the human behind the work. Shashir, what did you see in how Grammarly approached context that stood out to you and why did that make the idea of bringing Grammarly and Coda together so compelling? That's a great question. So, maybe just for winding time, about a year ago, I received a call from one of our investors who said, you should go talk to the team at Grammarly, they're working on some really interesting things and you should hear more about it. To be honest, initially, I was a little skeptical. I didn't quite understand how Grammarly worked, but I took the meeting as a trusted friend, and I learned a lot. So first off I learned that Grammarly was a much more pervasive and larger business than I imagined. Grammarly serves over 40,000,000 daily active users. Grammarly is in revenue scale terms, hundreds of millions of dollars in revenue. The Grammarly product does over a 100,000,000, sorry, a 100,000,000,000 LLM calls per week, which would probably rank Grammarly as one of the top AI tools, if you were looking at it just that way. But probably the most important statistic is Grammarly is used and integrates with over a million unique applications. Those are web applications, desktop applications, and mobile applications. And the way to think about Grammarly is that what Grammarly is really doing is Grammarly serves as the original AI agent. And so this was the second main thing I learned was not only is Grammarly at scale, Grammarly is heavily misunderstood. And the reason it's misunderstood is people think Grammarly is about grammar. Yep. And that's a totally reasonable assumption when you think about the name of the product and so on. But actually the work we do with Grammar while very important is just the tip of the iceberg of what is actually the technology stack that represents Grammarly. So the real technology stack of Grammarly is the stack that allows us to integrate with all these other products. So the way to think about this is in over a million different products, we can observe what you're doing, We can annotate it in a way that's unobtrusive to you and to the application, and we can make changes on your behalf. We sometimes refer to this technology layer as the AI super highway. It's where we bring AI directly to where you work. Or as we like to say, we bring an AI agent directly to where you work. But to just give a simple analogy in that construct of a super highway, today there's only one car running on that highway. It's the one driven by your high school grammar teacher. And your high school grammar teacher turns out to be a very impactful car. And it's tens of millions of people and hundreds of millions of dollars in revenue, but it's a significant underutilization of this infrastructure. So what we learned is bringing AI right to where people work is incredibly valuable. But what we think is that we can expand that into a much broader highway, so anybody can build on that same infrastructure and build their own Grammarly. That's what we call Go, and I think we'll talk more about that in a bit. Yeah, yeah, that's awesome. So Rahul, how does that idea of AI that already understands you, evolve as you think about superhuman and a world with multiple agents working together across tools and not just one assistant helping with one task? Well, we already mentioned that we need AI native tools that have three pillars, And the real productivity came gains come when our AI tools are not just ubiquitous, I. E. They work where you do, not just proactive, I. E. They help without you having to ask, but they're also connected. They know what you know, and they can actually act on your behalf. And the reason that's so important is that work today is, for all of us, fragmented across a shocking number of tools. For example, let's say that your manager sends you an email with questions about an upcoming product launch, and they're asking to meet so they can continue the conversation. The issues and their statuses will live in your issue tracker, like Linear or GitHub. The real discussion though on why the last few issues have been so challenging, well, they probably live in your team chats, like Slack or Microsoft Teams. Requests from your larger customers are probably logged in your CRM, like Salesforce or HubSpot, Whereas feedback from your beta users, they're probably gonna live in some, usually, a homegrown database that, honestly, may not even have an API. And this is all still just gathering information. To actually schedule that meeting your manager asked for, you're, of course, going to have to switch to your calendar and find times you're both free and now you're no longer in your email. So today responding to that email means you have to stop what you're doing and spend serious time in five plus other tools. So instead, imagine an agent that knows what you know across all of these tools and which can actually act on your behalf. In an instant, it would retrieve the information. It would write a full email that sounds like you, and it would schedule a meeting when you're both free in an instant. And again, that's when these AI tools start to feel less like a tool and more like a teammate, a coworker, a collaborator that we're used to working with every single day. Yeah, absolutely. You know, and I'm listening to you talk and I'm nodding and I'm thinking about all the different places that I go on a daily basis to retrieve information. I'm sure our audience is sitting there nodding as well. So, hearing about, you know, solving for that is music to my ears. Love it. Exactly so many. We just take it for granted now. Most people are even blind to this problem of switching. And by the way, every time we switch tasks, this is a study that has been replicated in human computer interaction multiple times, it takes our brain about twenty one minutes on average to recover from a context switch. That's every single time. So imagine what it's doing, switching context five times just to send one email, for example. Yeah, it's not good from a productivity standpoint, that's for sure. So what comes through clearly is that context only works if AI can actually act on it. And that's where I think most organizations get stuck. So even when AI understands the user, it's still trapped in these disconnected tools that we're talking about, in the tabs and the workflows, which brings us very nicely to our next trend, which is about AI Integration. Shashira, this is where your perspective really comes in. AI has exploded into a crowded and confusing landscape of tools. How do leaders think about, how should they think about the different roles that AI can play and understand what actually drives impact? Yeah, I think there's a lot happening in AI right now, and I know it can feel really confusing. And so I find it helpful to bucket some of the different trends and tools we're seeing into a few clear areas. So if you think about all the scaled AI tools, the ones that are horizontally applicable and have sort of scaled to tens of millions of users, they tend to have adopted one of three metaphors. So we sometimes call that assist, chat or do. So maybe I'll start with the middle one, chat. That is clearly the metaphor that most people introduced to today's AI have fallen in love with first. It was popularized by the product with the same name, ChatGPT, and now there are dozens of other chat tools and they all have the same metaphor. You should think of AI as a virtual human that you can go to and ask a question anytime you like. Products are very popular, but they do represent as Rahul described it, an AI detour. They represent a step outside your workflow. There's another track of AI tools that are very focused on what we call do. They're very focused on taking action with AI. For them, the metaphor feels closer to a task list. I'm taking an assistant, I'm handing them a task list, and saying go off and work on this and come back to me when you're ready. There's a lot of tools working on this, one of the more popular ones is the team at Anthropic has talked about how they've taken Claude in this direction. So Claude, the stat they gave is that about 39% of the query volume they're getting now is for these headless agentic tasks. Most of those going into the coding workflows with Claude code, but it's definitely a set of folks working on, think of AI as someone you can assign tasks to. There is at the bottom of that, if you were picturing this as a pyramid, there's a bottom of that pyramid that is the assist layer. And this is AI that comes to you. It's someone that's sitting on your shoulder next to you while you're working. And this is the metaphor that Grammarly has really focused on. And I think it's a very powerful metaphor. Interestingly, not a lot of other companies have this metaphor, but one that's very important to us. I'll give you one stat that really helps contextualize this. So I mentioned earlier, this fun fact about Grammarly is that we do over 100,000,000,000 LLM calls per week. We do that across about 40,000,000 users, which works out to a few thousand calls per day. And I just want you to think a little bit about what that means. So if you are a really good user of a chat tool or what we call a do tool, If you used it a dozen times a day, that would be a lot. But for Grammarly, you are probably, if you're a Grammarly user, we are using AI on your behalf thousands of times a day. And the way to think about that is every time you type a character, every time you switch to a new tab, switch to a new application, we are running a dozen different calls to figure out how best we can assist you on your task. So I think these are different metaphor for how to think about AI, but we've really focused in and we're doing elements of all of them, but I think there's an enormous opportunity in AI that actually comes to you. Well, what I love about that and the way that I personally, as a customer, have always thought of Crimely is just how you described it. You know, I've just got this little invisible angel sitting on my shoulder watching everything I do and making everything I do better. Expanding that conceptually beyond just, you know, a writing assistant and task assistants and things like that, that's really exciting stuff. That's right. And it probably leads well into what we're doing with Go. Yeah, absolutely. I was going to ask about that now. So how does this thinking, how does the thinking show up in what Superhuman is building with Go? And will you explain to us what an AI native workflow looks like when AI can actually take that action across tools? Yeah, and I think as we think about the different pillars Rahul described earlier, ubiquitous, proactive and connected, this is really where the element of connected comes out. But here's the way to think about Go. So we looked at Grammarly and we said Grammarly is really two different products. It's a product that we call this AI super highway, brings AI agents wherever you work. And then we have this very specific one we call Grammarly, that's the one that represents your high school grammar teacher. And so Go is really productizing the platform. It enables anybody to build agents that work just like Grammarly. They work right alongside where you work. So maybe just to give a quick example, if you picture Grammarly as like you said, the angel sitting on your shoulder that happens to be your high school grammar teacher. Imagine now that you're sitting down to write an email, and let's say you're a salesperson. So you're very lucky to have your grammar teacher sitting on one shoulder, but you also have your sales agent that has read and knows everything happening in your CRM. You have your support agent that knows everything happening in your support tools. You have your calendar agent that like your executive assistant knows everything happening in your life and your time. And now as you're sitting to write these emails, they all help you. One says, hey, you missed this key grammar pattern, you should reward it to be better understood. And then your sales agent comes in and says, you might be recommending the wrong product to this customer. We've actually found more success with this other product. And maybe your support agent comes in and says, did you know this person you're emailing had a major incident just last night? And it was resolved and you can acknowledge the incident and confirm for them that the incident has been resolved. And maybe your calendar agent can come in and say, hey, you're proposing to meet this person tomorrow at 7PM, but actually you have your daughter's recital at that time, and you should probably propose a different time. And so our aspiration with Go is to take that base platform of Grammarly and open it up so that anyone can build agents that work on top of this. Now we're building a set of those agents ourselves, and we've now shipped dozens of them that people can use. And we're also opening up that platform, anybody, whether a company or an individual, can make agents, maybe just for themselves, or to publish to the whole world, that can similarly follow this idea of being, ubiquitous, proactive and connected. I will confess that I will be signing up for that as soon as this webinar is done. Great! Because it sounds awesome! Alright, that brings us to our third trend: AI native workflows. Not layering AI on top of work, but redesigning work itself around AI as a collaborator from the very start. Removal in the report you used Superhuman Male's Story as an example of redesigning work around a big friction point. What parallels do you see between the email problem that you solve for and today's broader AI problem? Before we talk about mail specifically, I think the most important thing that we all need to do as leaders and as practitioners and people who are pushing organizations along is to revisit every process and every workflow and to redesign them, to use AI native tools. And I I say AI native as a shorthand for tools that are ubiquitous, are proactive, that are connected, that have really been rebuilt from the ground up to leverage the very latest in AI. And this is not simply taking an existing tool and bolting on an AI assistant or sidebar like Copilot or Gemini can feel it. It really means actually redesigning these tools from the ground up, whether that's done in house or you're buying these tools from some third party vendor. And today, most advanced and some of the most successful examples or at least the most well known are in engineering. Today, the best software engineers no longer write much code. Instead, they manage multiple AI agents. Some of those agents will be fixing bugs. Some will be refactoring the code base, and some will be working on large flagship features. And all of this work, this part's very important, will be happening in parallel. So AI native engineers are now 10 to a 100 times more productive than before. And soon, we're gonna see that in every function, in every department, and in every industry. In fact, we can redesign the entire productivity suite to be AI native, and that's a huge part of what we're doing here at Superhuman. So you asked about email and and what are the parallels between that and email? This is obviously a massive task, so we actually approached this in three increasingly ambitious phases. In phase one, which we started probably three years ago now, we built AI features that were on demand. Now these are features that you have to remember to use. An example from superhuman mail would be write with AI. You jot down a few notes, and then we turn it into a fully written email. These features are an easy place to start because they're only running when they are used. You don't need gigantic infrastructure, they don't cost very much to run. When we had success with phase one, we moved on to phase two, which is building AI features that are always on. So we've gone from on demand to always on. These are features that you don't have to remember to use. Examples would be things like auto summarize, instant reply, auto reminders, auto drafts, auto labels. There's a lot of auto in these that are just always working all the time. And these features are quite a bit harder to build, because even if a user isn't directly engaging with them, they require a pretty big build out of infrastructure. They cost real money to run because they're always on there, always working for you. And so far, we've processed many billions of emails through features like those. And now we're in phase three, which is where we're building AI agents that replace entire workflows. Imagine an email agent that triages on your behalf, that schedules on your behalf, that writes and even sends fully written emails on your behalf. An agent that can reason, that can problem solve, that can plan and create subtasks, that can incorporate deep context about you and your team, that can resolve ambiguity by asking you questions. It can go to the web. It can gather information. It can access other systems and tools. It can even query other agents. With such a system, you could imagine waking up to an inbox where every email already had a draft reply. You would simply edit and then send. And soon, you wouldn't even have to edit at all. That's my dream. That's what we're working towards. That's my dream. Well, trust and productivity often go hand in hand. And talking about this, thinking about getting up and checking my email and having all of that email already triaged and responses ready to go and all that sort of thing requires a great deal of trust. So when do you think we will reach a point where people are comfortable letting AI act on their behalf? I think this is such an interesting question because it is already happening. For tasks with low to medium complexity and stakes, we are in many cases, I think, already comfortable with AI acting on our behalf. And for tasks with high complexity or high stakes, we often ask the AI to make a recommendation, which we then review and edit. So as I mentioned earlier, the best software engineers today, they're not writing anywhere near as much code as they used to. Some things, however, are done completely by AI, such as fixing small gut small bugs. Whereas other things, they're done by AI and then reviewed and edited, such as producing large features. And we see this pattern in many other domains as well. So for example, in superhuman mail, we have AI that automatically triages your inbox. It hides marketing and cold pictures, the the kind that seem like real emails, you know, the clutter that even Gmail and Outlook doesn't hide, but we hide them. So you can focus on what matters most. And our customers are already so comfortable with this kind of AI working on their behalf. Most of them don't even review the AI working. That there's just trust that it does the right thing. Now we also have AI, like you pointed out, that automatically writes full drafts in your own voice and tone. And writing emails is clearly much more complicated than triaging emails, but it actually has fairly low stakes. You can review an email before you send it. So today, we draw the line right there. Just like engineers are reviewing and editing the output of a coding agent before they commit code, our users are reviewing and editing the AI drafts before they hit send. I will add that this is all changing very quickly, and I think our customers will soon trust us to send at least some emails automatically. Because some emails are less complex and they have lower stakes. For example, scheduling emails are simpler to write, it turns out, and also lower stakes than decision emails. And I would imagine that users would soon trust us to send scheduling emails automatically, especially if those emails were internal. I like it. You know, on the trust front, I will tell you my Grammarly story very quickly. Grammarly has been around for what, fifteen ish years? I was never a Grammarly. Seventeen, sixteen and a half. Did you say eighteen and a half? Sixteen and a half. Sixteen and half, okay, was close. I write for a living, I always have. I love writing. I love grammar. I am a word nerd and, I've never been a Grammarly user. And when I started, when I started working with Grammarly a few years ago, I thought, I'm not what I think a typical Grammarly user is, but I'm working with this company, I need to use their product, and Grammarly proved me wrong. And from day one, this person who thought, I don't need a writing assistant, I am an amazing writer, I am an amazing grammarian, Grammarly very quickly showed me that there was a better way to say things and there was a better way, a more succinct way to write things. Grammarly very quickly learned my voice and as I said, I quickly got to the point where I viewed Grammarly as this little angel sitting on my shoulder watching what I was doing, thinking about whether it was a text message or some other channel or something else, and just making everything that I did more polished, more succinct, more clear, all of those things. And so when you extrapolate that out and you think about, and I know that when you're listening to this and you're thinking about letting an agent have a certain amount of autonomy, that's a scary leap of faith. But when I think about this, I think back to my initial skepticism before I became a Grammarly user and how quickly I became a believer and how quickly I trusted in the solution, in the platform and everything else taking this additional leap of faith, which there is still this human in the loop and that's you and all of that sort of thing. But I think that, you know, that trust is a very big part of the equation and I think that in a whole Grammarly has done a very good job of winning the trust of customers, so taking these next leaps shouldn't be incredibly difficult. One, first off, oh, go ahead. I was going to say I'm going get t shirts made and say word nerd. So, have we're to work on that. I like it. Did you have something Rahul? Yeah. I had exactly the same experience, which is, you know, English is my first language. I, for better for a word, a lot of writing training and just grew up and had spent most of my professional career before Grammarly slash superhuman believing I did not need AI writing assistance. And then, of course, here, I I wanna use all the products. We're looking for interesting ways to integrate them and and make them work as an amazing, incredible product suite. So I do use Grammarly every day. And you're completely right. It continually blows me away by how there is still room to improve what I would organically write. And how despite despite having actually, I think, pretty good spelling and at least what a a UK grammar school, that's actually what we call them in The UK, grammar schools Yeah. Yeah. Would consider perfect grammar, there are still room to improve. Well, and for me, was very much of a learning experience and I still find, you know, I've been using it for years now, but I still find myself stepping back and going, you know, that is better. Know, one comment on this discussion that I was with someone a couple of days ago who is a book author, and was describing a similar, he was writing a second book and he said, the first one I did was just an editor, and then I had Grammarly and I got, it turned out I needed my editor a lot less. And then he started talking about his dream, and we talked a little bit about what we're doing with Go, and said, know, up till now, the epiphany you went through of actually having my grammar angel on my shoulder is really meaningful, was really amazing. And so what angels would you want to have on your shoulder? And as he was describing it, he started dreaming and saying, you know what, what I really want, everybody who's writing a book has another author they love. Like, my book could only be as good as this other author. Imagine you could hire that author and you could say, can you sit on my shoulder and just, I don't need you to correct my grammar. I need you to make sure what I write is as compelling as what you do. What would that feel like? One of the examples, one of the first go agents is from an author named Kim Scott, who wrote a book called Radical Candor, and that's exactly what she did. She built an agent that takes her philosophy and turns it into something that sits on your shoulder and helps you follow the philosophies of radical candor. So I think the idea, and I often hear stories from people about, I was surprised how much I needed Grammarly and so on. And then I'll ask them, okay, let's ignore grammar for a moment. Who else would you want sitting next to you, helping you in everything that you do? Yeah. And for that, the list is infinite. There's just so many ideas of ways that we can help people that I'm sure not we haven't even thought of most of them, but I'm really excited to go down that path. Yeah, well, I'm excited for you to go down that path too. Get going! Alright, gentlemen, it's now time for audience Q and A. We have gotten so many amazing questions, so thank you everyone for writing in. I'm going to group a few themes together so that we can give answers that are broadly useful. Shashir, the report paints a picture of AI that is ubiquitous, proactive and connected. Which of these qualities do you think will drive the biggest impact? Think it's a great question, and they obviously form three legs of a stool, so it's hard for them to work without the other. But if you were to think about each of them, and use a little bit of a human metaphor, so what do each of them mean? So when we say ubiquitous, proactive and connected, I often think about it as imagine you could hire a team of people, what would you want them to do? Ubiquiti says, I'd want them to be where I am all the time. Connected means, I want them to know what I know. I want them to have read everything I've done, read my documents, read my email, know my CRM, what's happening and all the things that are important to me. And proactive means, I don't want them to sit on the side of the room. I want them to interrupt me in a way that is as least disruptive as possible when they have something that would be worthy of my attention. So I think each of those is very important. Proactivity is definitely the start. It's very hard to do the rest if the person isn't even in the room. I think connected takes someone that is perhaps generically helpful and makes them specifically helpful. Makes them understand me, makes them understand my data, my context and my systems. And proactive is what avoids the AID tour, where we started the conversation. If I have to remember to go to this person, you get a very different impact. And I think one way I often reflect on this is, I remember when I first worked with an executive assistant, a human executive assistant, and it took time to understand how to work with them and how to work them into my workflow. In this new world of AI, you don't have to do that. You can build something that, or work with tools, that achieve all three goals of being ubiquitous, proactive and connected. But I think they sort of stack in that order. Yeah, yeah, I agree. I'm still thinking about the connectivity part of this. None of these three are more important than the other, but I think that connectivity part of it is so much of what we get lost in, in that productivity paradox, is that pinging from place to place and that sort of thing. So being connected to every place that I'm connected and every place that I'm working so that I don't have to remember to go do these things and things like that, that to me is just such an important part of this equation. Absolutely. Alright Rahul, here's one for you. How should leaders decide which workflows actually need AI and where restraint is critical to perhaps avoid adding complexity? Okay. So this may seem extreme, but I think that leaders should be looking to maximize AI usage in every single part of their organizations. And things that AI can do today, well, they were simply not possible as recently as six months ago. And, that time frame also seems to be compressing. The not possible time frame used to be like two years and then one year. Now it's six months. I I wouldn't be surprised if it becomes four months, five months. So whenever we think AI can't possibly do this or, yeah, we tried that a year ago and it didn't work. I think we have to check ourselves. We have to reassess reassess our priors and we have to think again. And I don't mean to sound alarmist, but if you don't do this, your competitors will. I really do believe that the only way to stay competitive in an environment like this is to make AI transformation a top three priority. What I would suggest restraint or at the very least triage is to start with the highest leverage areas. And I think one of the highest leverage areas are the horizontal tools like that we all use every day. So, writing, email, chat, calendar, documents, meetings. I I just described the the full set of hours of a modern day professionals working, life. And and that's what exactly why we're building all the tools that we're building. That's why we're building the AI native productivity suite of choice. So restraint, Definitely not. Go full maximalist. Apply AI everywhere. But really think about starting with the horizontal tools that affect everybody in the organization. Yeah. Yeah. And I will add, apply AI everywhere and make it available to everyone across the board. And I say this because I've had many, many conversations and I've done many interviews with, executives charged with leading AI initiatives in enterprise organizations and across the board, the ones who are having the most success are the people who open up AI to everyone in the organization and that's really when the magic happens. Absolutely. Yeah, so Shashira, let's talk change management. Here's a question about what role does change management play in solving the AI productivity paradox and where do most organizations underestimate it? What a fun question. Maybe I'll take it on a little bit of a tangent for a second and recommend a book. So, I have a number of different books I recommend, but whenever I'm asked about my top five recommendations, this book takes two slots in the top five. So it's a book called Switch, it's by Chip and Dan Heath, and it's about how to change when change is hard. It's a fantastic book, highly recommend reading it, I reread it basically every year. And it's worth reading the book, but I'll give away the main premise. So the main premise is that when you're trying to cause change, they use this metaphor of an elephant on a rider on a path. Sorry, a rider on an elephant on a path. I can get the order right. A rider on an elephant on a path. And you have three things you can do to cause change. You can direct the rider, you can give them instructions, you can motivate the elephant, You can give the elephant energy. You're not exactly sure where it's going to go, but once the elephant's moving, it's going to move. Or you can construct the path. You can construct boundaries such that the elephant can only go a certain direction. So in that way, that book walks through many different examples of trying to cause change. It might be in your workplace, it might be in your personal life, it might be in your industry. And they talk about in every case, you can take the patterns and they fall in one of these three buckets. You're either directing the writer, you're motivating the elephant, or you're shaping the path. So as we think about change management with AI, I think that the three buckets directly apply. Maybe I'll start with the elephant. So the number one thing you have to do when driving change management, the way Rahul just described, of really working without restraint to get AI deployed in your organization as fast as possible. The first thing you need to do is convince the elephant to be excited and not to be afraid. It's the fear versus the dream. And I think that unfortunately, there's been a lot of rhetoric about AI that leads to fear. Sure. And I think, there's people scared of it in many ways, they're scared of job replacement, they're scared of hallucinations, they're scared of many different things. And I think it's often hard for them to dream the right way. And for that, I think that the right metaphor is very important. I think if you think about AI as a human replacement, I think you get to very, very different ways of thinking about and being afraid of where I will go. But I think in most cases, the best way to think about AI is as team augmentation. It's the angels on your shoulder. It's the fact that you can have one, two, three, ten, 28 angels on your shoulder, helping you be you. You're in control, you're the one that's making the choices, they can do meaningful parts of the work with you, for you, they are ubiquitous wherever you are, they are proactive in bringing things to you and they're connected, they know what you know, but it's still you. And so I think first thing I would say is as you paint the dream for your organization, it's really important to just get the metaphors right. The second I think is direct the writer. And direct the writer is really about, you have to sometimes give instruction, people don't know what to do. And I highly recommend working it in to your next off-site, into your next staff meeting, run the exercise. What could it feel like? Stretch people's views of what they might be able to do. One of our favorite examples was I watched our recruiting team take a set of AI tools and say, we're going to recommend product features. Our recruiting team doesn't work anywhere near our engineers. Went and produced actual live prototypes of things they wanted in the product. And they were able to do it and it changed how they thought about it. So direct the writer is very important, get them on that path. And then that last part is the path itself. How do you shape a path that makes it very hard not to be AI native? And the key for that is choosing the right tool set for your company. And I think it's very important that as you evaluate tool sets, and I know there's a lot of them that are older tool sets that are trying to bolt AI on, but that idea of picking tools that are AI native, that are thought from the ground up in this way that thinks about AI as embedded directly in the product, making it ubiquitous, proactive and connected. And I think what Rahul said is very important. A lot of the AI tools are the specialized ones, but actually the horizontal ones are the ones with the biggest impact. It's where you spend all your time, whether you're writing or communicating or collaborating, those are all the areas where you actually spend all the real time working with other humans, that's when your agents can be most helpful. So I think if you want to motivate the elephant, help them feel the dream, not the fear, Direct the rider, work it into your next exercises, and then shape the path. Make it easy. Pick the right tools, set them up so that they're already ready to go, make sure they're connected to all your different systems, and your employees will naturally start adopting them without even thinking about it. Excellent. Switch. I'm officially adding that to my must read list. I've also got, I've got twins who are college sophomores, and, I think I'm going to put a copy in their hands because if anything, they need to learn about things like this early on in their careers. Very readable book. I appreciate that recommendation. So here's another one, this is for you Rahul, how can leaders move from isolated AI tools to those deeply integrated systems that we need without completely overwhelming their teams? Let's see. I think it's about figuring out what soft skills folks need and what hard skills folks need, and then helping people develop and adopt those. So on the soft skills side, I think about adaptability, flexibility, ambition, courage, a growth mindset. These are all things that you can also approach using the rider path elephant metaphor. It's another way of looking at a similar set of issues. And I think actually the best organizations are actually selecting for and developing these things, whether or not they're AI native. In fact, you could argue the causality is the other way, that the best organizations are the best precisely because their people have these soft skills. And if anything, it's this question is a good reminder that amazing outcomes start with amazing teams. And what we really should be doing is making sure our our teams are incredible, and we're giving them the environments to to blossom and and really achieve their life's best work. But there are also hard skills around using AI. Those are things like understanding the capabilities of the latest models, keeping pace with the latest tools, and then communicating these things internally. So I think as leaders, we need to do two things. I think number one, that we need to make it very clear that these hard skills are expected. This is instructing the rider, and that it will be well rewarded. I'm an optimist, generally. I think that the best people are going to rise to that occasion. And number two, I think we also need to give people room to develop these hard skills. For example, we can't simply ask our engineering teams to become AI native and expect it to happen overnight. There's a lot of trial and error involved. Misha Chao mentioned choosing the right tool set. You have to do that, of course. But then you also have to give people room to experiment, try, and fail. Failure has to be acceptable. And there needs to be time and space to actually figure out how these hard skills work. Because we're all still figuring this out. Like, I don't think there are definitely some companies that are ahead, but generally, most people don't have the answers. There isn't a silver bullet. So we have to invest in the process and the tooling to make it possible. You know, I couldn't ask for a better way to wrap this conversation. That was fantastic. So this brings us to the end of today's discussion. Before we wrap, we've just launched a poll that you should now see on your screens. Let us know if you would like to keep the conversation going with one of our product experts. Thank you so much to Shashir and Rahul for such an engaging conversation and thank you to everyone who joined us live. We hope this session gave you some practical ideas for how to approach the AI productivity paradox and to start designing truly AI native ways of working. A recording of today's webinar, along with a link to the twenty twenty six AI Shortlist, will be shared with you in a follow-up email. And again, thank you so much for spending time with us today. Have a great rest of your day! Thank you. Take care.