How Zapier’s EA built an army of AI interns to automate meeting prep, strengthen team culture, and scale internal alignment | Cortney Hickey

[00:00] I think from EAs I hear like, oh, AI doesn't think the way I do. I'm like, it can though. It can. As long as you can figure out the system behind why you're doing things and be able to articulate that. I think this is going to be one of the most practical, time-saving, and stress-saving episodes of how IAI we have ever had. This agent does everything you want and then more, and over time you can make it more intelligent. It's serving in this kind of second brain for me where, yes, [00:30] of my superpower as an ea is remembering all these things about people but this is making sure that it's not all in my head and i can really refresh my memory quickly on all the context rather than diving through the crm my email slack and looking at all these things separately if you are doing a repeated task every week spend time that week automating that task i definitely am a believer that ai can only enable us in this role i think it's a when not an if we will have to be folks [01:00] tools. Welcome back to How I AI. I'm Clara Vo, product leader and AI obsessive here on a mission to help you build better with these new tools. Today we have Courtney Hickey, EA to the CEO at Zapier. And yes, of course, [01:17] She uses AI to automate all of the admin tasks related to meetings and document preps and feedback [01:24] But she's also going to show us some unique ways that you can use AI to reinforce your cultural values and operating principles. This is a really great one for anybody thinking about organization at scale,

[01:36] operations at scale, and culture at scale. Let's get to it. This episode is brought to you by WorkOS. AI has already changed how we work. Tools are helping teams write better code, analyze customer data, and even handle support tickets automatically. [01:51] But there's a catch. [01:53] These tools only work well when they have deep access to company systems. Your copilot needs to see your entire code base. Your chatbot needs to search across internal docs. And for enterprise buyers, that raises serious security concerns. That's why these apps face intense IT scrutiny from day one. To pass, they need secure authentication, access controls, audit logs, the whole suite of enterprise features. [02:18] Building all that from scratch? It's a massive lift. That's where WorkOS comes in. WorkOS gives you drop-in APIs for enterprise features, so your app can become enterprise-ready and scale up market faster. Think of it like Stripe for enterprise features. [02:48] today. [02:49] Courtney, welcome to How I AI. I am [02:54] Really excited about this episode because I think this is going to be one of the most [02:59] practical. [03:01] time-saving and stress-saving episodes of How I AI We Have

[03:07] ever had. So I'm just pumped to have you on. Can you just tell us a little bit about why you've chosen to dive headfirst into using AI in your role aside from the place? [03:20] that you work? Yeah, so I work for Zapier, which is an automation and AI orchestration company. So [03:27] Of course, it's part of our company ethos, but I am just personally super passionate about using AI because I think it can help. [03:38] work myself out of the boring, repetitive manual parts of my role so I can do more interesting work. And so I truly believe that it's not a if you have to use AI in this type of role, it's a when. So I like to be ahead of the curve. I like to learn by doing. And so I've spent as much time as I can over the past couple of years really diving into this and seeing how I can change the shape of my role with this new tech. [04:04] And what I like is we're going to start off on a workflow and use case that I think everyone can relate to, which is... [04:12] meetings stink or not meeting stink, but meetings could be better used in most organizations. They're expensive. You have a lot of people in them. [04:23] And I think like prep and follow up are so valuable and aren't really done well by organization. So I'd love for you to walk us through a couple of your meeting related workflows. Totally. Yeah. I mean, as an EA, my life runs off of the calendar. So that was naturally one of the first place I dove into with AI.

[04:43] And so let's jump into one of my favorite workloads that I've built, and this is within Zapier Agents. So our Agents product, within Zapier, we have a bunch of different products all the way from... [04:57] super deterministic automations that run the same way every time with little creativity to these agents that can do [05:05] tasks that involve more reasoning and have a lot more freedom to operate. So you could build this in a zap, but I wanted to like, [05:15] paint the, you know, change the color of the sky with this agent for myself. So this is an example of one I use personally, but you can replicate this for [05:22] anyone you work with, but [05:24] This is essentially my weekly meeting prep. [05:27] which on Fridays I used to have [05:30] maybe let's say two hours blocked, 30 minutes to like do a retro on that week and anything I need to change, but then like spending an hour or so really diving into the next week and what I need to be prepared for. [05:43] The way this agent works, and it's kind of developed over time, but it has a few steps it goes through. And the key thing is this is scheduled to trigger every week. I have it due at Friday 8 a.m. So you can have it whenever. And basically it goes through my calendar for the upcoming week. It identifies all meetings that require prep. So I'm going to go through it. [06:03] Personally, I don't need to prep much for routine one-on-ones, routine stand-ups, or recurring internal meetings, but I do have more and more external meetings on my calendar now that I'm doing more out in the world with AI and automation and teaching folks how to do the same. So...

[06:16] Basically, I have it be a bit of a research buddy at first. So first it just pulls my calendar, then it goes and does all the research for me. So [06:23] This takes anyone without a Zapier email and does a web search, basically. It researches their current role, their industry experience, anything noteworthy that I might want to know about them. And then it does this cool thing, which also goes and checks in our CRM, which we use HubSpot at Zapier. [06:41] You can put in any of our 8,000 integrations here and use your CRM, but [06:47] For each external participant, it goes and then looks at what their relationship is with Zapier. [06:52] It looks at email address first, then my company name, figures out if they're in a deal, if there's any recent sales team notes, if there's any interaction I should know of. [07:01] And then it also goes and searches internal comms history. So within my Gmail to see our prior relationship, if any, within Slack to see if there's any call outs to their company. And so it's doing all these things that I would do manually. [07:16] then it's delivering me two outputs in the end. So one of them is [07:20] tasks for my [07:23] actual prep with all this in it. So I like it to create a task in Todoist, which is my to-do list app within a certain project of meeting prep. It pulls in all of this information with this intelligence from the agent and tells me to prep for it. It puts it on my calendar for two hours before the meeting start time. So you can see what that looks like. [07:43] in real life you know here here's a couple meetings i have next week and they're automatically

[07:47] cued in my Todoist [07:49] but [07:51] The second thing it does is delivers this weekly digest to Slack. So [07:55] This is, and you could do this day by day too if you have a ton of meetings, but again, I'm mostly internal. So [08:02] I have it create a structured digest, which includes all of the meetings and intelligence, um, [08:08] Any error handling I might need to know, like if the agent couldn't find someone or if there's anywhere I should do a manual follow-up. [08:15] And then it does this like [08:17] It pulls its own, uses its own creativity to create these insights for me about [08:24] what I should pay most attention to for this upcoming week. So I can pull up a [08:29] A real example here. So I just ran this this morning just for this example, but it's going looking at the next week. It's pulling all of the key meetings that require prep preparation. [08:41] It's saying, okay, so we have a team onboarding with Fellow. We're changing our AI note-taking app. And so it did some [08:48] intelligence on who we're meeting with for this. It's confirming that this is a new vendor relationship. We've previously purchased this meeting management. They're our implementation specialist. [09:00] And then, you know, here's here's more context on this. So it's like it's it's serving in this kind of second brain for me where, yes, I have all this. And part of my superpower as an EA is remembering all these things about people. But this is making sure that I don't it's not all in my head. And I can really refresh my memory quickly on all the context rather than diving through the CRM, my email, Slack and looking at all these things separately. So one place is key. And then.

[09:29] It pulls in this like key prep recommendations at the end, which is where the agent gets a little bit creative here. So... [09:36] It's saying I should review my previous CARV session. This is an EA automation session I do every once in a while. [09:42] tells me to premiere some new demos, [09:44] Tell me to familiarize myself with Fellow before the onboarding session and check with our head of marketing for PR priorities before the agency call. So I love that it, you know, does exactly what I need to do. Like, you know, gives me all these preps in my to-do list and does those actions. But it also kind of serves as like a double check. Maybe there's something I haven't thought of. Maybe I didn't think I needed to update my deck and it gives me something new. So I think what's great about this agent is it does everything you want and then more. [10:14] make it more intelligent. So as you learn how this works, and so... [10:21] I'll give you an example of how this actually works in reality. So this is the test that I pulled right before this call just to give us a clean Slack output. And it walks you through step by step what this agent was thinking. It's like, okay, I'm testing this. [10:34] I'm going to go look at the calendar. I'm going to go research all these participants. You can click in. [10:39] and see even more information about what it was thinking, [10:43] You can see that it went in HubSpot, couldn't find someone for there, couldn't find someone. We probably didn't have a relationship with them. [10:49] Oh, great. I found someone. [10:51] And then it tells you everything it did. So over time, if something's not performing as you intended or you want to update it, [10:58] you can really look at how this agent works on the back end and give it some feedback and

[11:03] We have this great co-pilot where you can go in and say, [11:07] You could say like, [11:09] You could go into Copal and be like, "Oh, I actually would love to have a hyperlink to their LinkedIn page included in my Todoist thing." So you can say, "For each participant." And you also... [11:26] add a LinkedIn hyperlink. [11:29] within the Slack digest. So you can kind of [11:34] I always tell people when they're starting with an agent like this is, [11:37] Progress over perfection. Like I started this one with just a quick digest. It didn't have our CRM connected. It didn't have that. [11:44] And then over time, I was like, oh, here's something else that might be helpful. And so, like, build something basic, see how it works, learn, and then, you know, make time to improve it over time so that you make sure it's really being impactful for you and doing work. [11:59] all of the things it can. And, you know, these tools are getting smarter every day. So also keep on top of, you know, the new... [12:06] new capabilities so you can start building those into agents and automations and things that you've built in the past. So that's a quick overview of the engine. [12:16] Something I want to call out for folks is I think this workflow highlights a couple strategies that I think people really need to think about. [12:25] One is [12:27] I tell people if you are doing a repeated task every week, [12:32] spend time that week

[12:35] automating that task. And so I, when I had fancy jobs, had an EA as well. We had a very similar process where on Fridays, we would actually do a retro the past week, [12:46] "Prep for the next week, find out all the stuff we needed to prep, make sure that I knew everything that needed to happen." [12:52] And instead of spending that hour [12:55] doing that prep on a Friday, I highly recommend people just say this, this week, I'm going to spend an hour automate, automate agent, and see if I can replace that flow. And so I think that's a really useful mindset. [13:09] to bring into [13:11] what and when you can automate. - Yeah. [13:14] The second thing I would say is I love agents in particular, the sort of like natural language format of describing agents, because you can literally just narrate what you would do. You would be like, first, I would go to Google and I would look at all the meetings for the next week and I would decide which ones. [13:31] I need to prep for. [13:34] Then I would go look at my email and see what the heck we're actually meeting about. Then I would dig through Slack. [13:40] I would probably go look at HubSpot. And then I would, if I was doing a great job, organize it in this way, send it to myself in Slack as a reminder and create a bunch of to-dos. Yeah. You can actually use that natural language to... [13:54] describe an agent structure. And so I think it's a really natural way for people to get started designing some of these workflows. [14:01] Yeah, I agree. Like I think of agents, when I first started using them, I

[14:05] kind of started thinking of them as interns almost. So they're not going to operate and do something completely independently from the start. But if you can teach the intern your system, and how you think and give it the tools it needs, then over time, your intern gets smart enough to to run and do things on their own. And so, you know, this is something that now I rarely touch this agent, because it works as I planned consistently. You know, right now, that was a [14:35] LinkedIn profile so I can quickly add them. But, you know, there's not much else I have to do here. And now I've given myself that time back. And even bigger, I can showcase this to people [14:47] everyone at Zapier, enable them with this template, which you can share, and then everyone can have this meeting prep agent. They can, you know, they can add different things if this isn't their exact workflow, like not everyone uses Todoist or, [15:00] you know, not everyone wants X, Y, Z, but they can customize it for their own. And so I think it's like, [15:06] Yeah, teach people how to fish and teach these interns your way of thinking, these agents, [15:12] over time you'll be surprised of how much you can do. I think from EAs I hear like, [15:16] oh, you know, AI doesn't think the way I do. I'm like, it can though. As long as you can figure out the system behind why you're doing things and be able to articulate that. But yeah, I love the, like, the dictate to co-pilot too, because I do that. I'm like, okay, so usually I talk to it just like that, like, as if I'm on a walk with a friend and see what it comes back with. And,

[15:39] So, yeah, this is like one of those things that's just... [15:42] a no-brainer to spend a little bit of time on and then it just runs in the background. [15:48] Yeah, and I think, you know, EAs in particular are so well positioned. [15:53] to make some of these tools for the broader organization because [15:58] you know, you're a point of leverage in a team. [16:01] And if you can systematize that leverage, I think two things happen. One, [16:07] you can do a higher level [16:10] job supporting your exec or your team to everybody else gets a little bit of a boost that you wouldn't be able to personally give them. And so yeah, I think, you know, [16:21] everybody should think like, oh my gosh, I could have my own little mini assistant or I could have my own little army of interns if I can just describe what I need them to do. And I think that's really interesting. The last thing I will say is I have a very almost exact workflow in Zapier Agents. It's called my Sunday Scaries Prep. I do it on Sundays when I start to feel lots of anxiety. Now what I'm planning for the next week. And the one ad that I put in here is [16:51] actually mix. [16:52] professional and personal stuff. So I put in there [16:56] If my mornings allow me to walk my kids to school, [17:00] Block awe. [17:02] you know, this hour to this hour, [17:04] because I know I can like walk the kids to school. And by Sunday, if you haven't booked me on an early morning, you don't get me. And so like add these little

[17:12] you know, call out days that I don't have time scheduled for lunch, like call out days where I have six hours of back to backs with no break, like, give me an opportunity to improve my calendar. So I do think in addition to prep, you can do a little like calendar optimization, too, which is really nice. Totally. I agree. Like, yeah, which meetings might be able to combine or get rid of that look duplicative, you know, give it give me some recommendations for optimizing [17:42] totally combined personal and professional calendars into this to make it a jack of all trades and do everything. But this one, yeah, this one for me is focus on work. [17:51] You know, if you really want hyper efficiency, you just make an agent that says, find all my meetings, cancel all of them, give me my day back. Yeah, the Ron Swanson agent, I don't know if you watch Parks and Recreation, but April Ludgate scheduled all of his meetings for like March 31st to one year because she didn't think that existed. Yeah, exactly. Perfect. Schedule all of my meetings for March 31st. You know, you have one other meeting related workflow, which I think is really interesting, which is making sure that the meetings that you do really are high value. [18:21] So I'd love for you to walk us through what you do there. Totally. So there's a few things on the other side of meetings that I do. So one of them is, you know, this is Wade, the CEO of Zapier. So I was basically, the way this workflow came up was we use Fathom for our meeting note-taking. So I was manually going into Fathom after each executive meeting and giving it a prompt,

[18:51] the team, which is a framework we use from the table group, or who in this meeting could have spoke up more. And I was giving it prompts to see how Fathom did with more reasoning and more of a loose like feedback, creative prompt versus what were the action items, which of course it does excellently. And over time, I was like, this is pretty useful. And so I was manually doing that and Fathom sending it to the exec team. And Wade sent me this message. He's like, I feel like we [19:21] pretty useful accountability mechanism. I think the other thing here is [19:24] when feedback is maybe [19:27] automated, growth feedback is one of our values, so it's part of how the company runs, but when feedback is [19:34] expected after a meeting and becomes a part of routine and coaching. [19:39] then folks learn to expect it and it's part of their behavior and it doesn't make their like, you know, that nervousness spike up when they get feedback come in. So I think the more feedback folks can get, the better. But... [19:50] I think the other thing this does is take some spit off the ball. So, you know, after meetings, I've worked at this team, this exec team for five years. So no one would... [20:00] be offended if I said like, [20:02] Brandon, you really should have spoken up on that topic. I can call them out because they've given me the permission to do that. But for folks who are newer to organization or don't have that comfort level with the team, you can build this meeting coaching across the org and automate it based on any meeting transcript. So I started working with him with Fellow because we're moving over to Fellow for AI note taking and was testing an agent. I was giving...

[20:30] Wade, you know, here's an example of the feedback it generated. It gave him speaking time. It gave him, you know, what went well. It gave him opportunities to amplify impact. And then he's like, this coaching is too soft. Like this, this is still what went well. Let's have it be tougher. So then I went. [20:47] give it you know i fixed the agent instructions to give it a better balance of being demanding and supportive which is a term we use a lot um here at [20:56] Zapier of like, you have to be a demanding leader, but you also have to be supportive. So I gave him this one, which did give some more growth opportunities, like address misalignment more directly, challenge the decision-making speed. It seems like you have some fear of conflict. And so we worked at that, gave it a more concise version, thought it was good enough to ship. And then so now we've shipped this kind of meeting feedback automation system through Fellow. But [21:26] meeting metadata, make sure there's some, you know, some parameters, like if a meeting is only 10 minutes, probably not worth the feedback and, and make sure it's only zap your employees, make sure there's, you know, only sufficient context to offer valuable specific feedback. And then for each participant, it can look up their Slack, match their email just to Slack, and then send them some, you know, some feedback. And I gave it context on our company values, you know, some of our meeting

[21:56] behaviors for like what we expect from folks at Zapier. And then I gave it the five dysfunctions of a team. And so I really worked on this prompt over time to help it generate this direct constructive feedback on all these dimensions. And then, you know, you can see what the outcome is going to be. And this is, you know, clarifies it's AI generated. It's coming from a bot. It gives [22:17] you know, very quick feedback after a meeting of one to two specific growth opportunities and one to two things they can do next time. So this is something that I think can... [22:28] over time really just change the way that a team works together and change the usefulness of a meeting. So [22:36] This is maybe not something that was [22:39] a huge part of my job and I don't calculate this as like a big time saving agent for myself, kind of like the meeting prep was. But this is something that's like really reinforcing the company culture and making folks better at their jobs. So I think this is it's cool to build stuff like this that's more. [22:57] That's more just enablement and accountability for folks, especially among the exec teams. You make sure that they're being like the best [23:06] displayerism of company values and norms over time. So this is a fun one that I had a good time creating with Wade. [23:13] This episode is brought to you by Brex. If you're listening to the show, you already know AI is changing how we work in real, practical ways. Brex is bringing that same power to finance. [23:26] brex is the intelligent finance platform built for founders with autonomous agents running in the background your finance stack basically runs itself cards are issues expenses are filed and fraud is stopped in real time without you having to think about it add brex's banking solution with a high yield treasury account and you've got a system that helps you spend smarter move faster

[23:56] runs on Brex. You can too at brex.com slash how I AI. [24:04] So Courtney, what I think is great about this is people really think that culture is hard to systematically reinforce. You really think that culture has to be something that [24:15] individuals or leaders have to carry through sort of soft interactions with the organization. [24:21] But what you're showing here is more than [24:24] Hey, can I give you, you know, skills coaching on closing a customer or [24:30] I give you [24:31] communication coaching on managing stakeholders. This is [24:35] are we embracing our [24:38] operating principles, our cultural norms, are we keeping an eye out for issues in interpersonal conflict or communication that we know are [24:47] teams are biased towards? And are we creating sort of a [24:52] ego protective system [24:55] in order to [24:57] continually check and keep ourselves [24:59] accountable to that system. And so I think like, [25:02] Take the meeting part of this aside, [25:05] the ability to [25:07] kind of consistently check interactions and projects and initiatives inside your organization [25:13] with alignment, [25:14] on your stated cultural norms is a really powerful thing. [25:20] thing. And you mentioned table group. I've worked with them before. And you get them like once a quarter. All this great work for your leaders. And you're like, yeah, we're going to be the best team. We're totally aligned. We're giving everybody feedback. But they're not whispering in your ear during the executive meeting. I mean, I'm sure they would for a price. But during your executive meetings, they're not listening into your company town halls or AMAs. And so I think this is just

[25:50] to observe your organization from a third party kind of like vantage point. Yeah. [25:55] And then as you said, just normalize feedback. This is very stressful feedback for people to receive maybe from their boss. Like, hey, you didn't do this or you didn't do that. But if you know everybody in the meeting... [26:08] Is getting feedback. [26:10] It's coming from sort of a neutral evaluation place. [26:14] then you might be more open to hearing and kind of adjusting your behaviors based on that feedback. [26:21] Yeah, totally. I think you hit the nail on the head with a few of the main reasons why I like this. I think [26:27] It's [26:28] Are we who we say we are? You know, are you who you say are we what we say our culture is? And are we keeping that top of mind in between things like, you know, we're a fully remote company. We do only meet with table group once a quarter, once a half. And so we need to keep these behaviors top of mind consistently. And. [26:46] Folks have a hard time keeping anything top of mind for that long. And so making sure this is repetitive and [26:52] Continuing to reinforce those things is valuable. But yeah, we've I mean, we've got the other type of agent for, you know, sales reps, for example, that gives them after gone calls what they could have done. That's more like, okay, you should have brought up this ROI or this metric or, you know, more specific sales coaching. But I love the culture stuff. [27:10] Okay, this is great. So we have schedule prep, we have culture checkpoints, which I think are awesome. But let's answer the question with AI that I'm not saying every IC manager and leader in the organization thinks about a lot, but they might, which is...

[27:28] "Will this fly?" [27:30] CEO or will this fly with executive A or how do I know I'm not walking into into a tough meeting so you've done you've done some work to sort of stress test [27:42] what your CEO [27:44] might want or participate in without having to bother him. So I'd love to see a sort of exact replicate. [27:52] workflows. [27:54] Totally. So yeah, again, on the meeting side, but this is a GPT I built within OpenAI's chart GPT. And it's we have this public [28:05] feed in Slack, which is feed T-ups, which are basically any strategic doc that need review across the company. We've kind of centralized this in a feed for transparency, accountability, and make sure folks know why we're making certain decisions. And so [28:18] The folks were often coming to me for thought partnership of like, hey, here's my doc. Do you have any feedback on it before I share this with exec? You know, you know how Wade thinks. You've worked with them for five years. What do you think about this? And I love doing that thought partnership stuff, and I don't want to replace it. But again, it's not scalable, and I don't want to be a bottleneck for someone to get their doc out into the world. [28:48] sharpen these tee-ups. And so... [28:51] Sometimes it makes sure that, you know, folks come into a meeting with more confidence and their opinions are stress tested and the right data is included and make sure we make the most of those meetings that we're in. But sometimes the tee up is so clear now that we can skip the meeting entirely, which is great as well. So you can see here, like, you know, Wade said he tried it out, caught several things that would strengthen the work. We've got, you know, Lindsay saying, I was worried it would just tear my dock apart, but it suggested really great simple tweaks.

[29:21] Thank you. [29:22] We gave it a bunch of knowledge. So I'll dive into... [29:26] that right now. So, um, [29:28] This is the exec prep GPT. [29:31] I... [29:31] Gave it a prompt here to say give feedback on a tee up doc, which was one of the main prompts considered and I created a fake doc Which is a very very poor tee up just to give an example for this so this was a tee up that I [29:45] Basically, I just said create a really bad EATF doc. So this is like, you know, gives a really loose purpose. You know, it doesn't have an approver. It's like... [29:56] Just something we've been thinking about doesn't have much background. So it's a bad doc. So it's not but it's not going to give you the most exciting feedback. But it goes in and again, like takes the spit off the ball of feedback and helps people, you know, get more confident. [30:12] This is saying, you know, this reads more like a jam session than a tee up. You will get better feedback if you clarify these things. Let's tighten it up. And so it gives a quick read of what you have, gives feedback on how to strengthen it, to surface tradeoffs, add a recommendation. [30:27] And then, you know, it gives an example rewrite even in this case because the doc was so loose on details. But it gives, you know, a couple top fixes before the bullpen and then, which is what we call these kind of T-Up meetings, and then one bold coaching question. So I love that it does this and even gives you a suggested next step. Like, let's give a tight Wade-style one-page rewrite of this T-Up so it's bullpen ready. So I love that.

[30:57] This GPT is built off of the back of, I'll dive into it really quick, but, [31:04] It's built off the back of, you know, [31:07] Again, our team norms, our revenue roadmap, our strategy memo, good examples of tee-ups, you know, wade feedback tuning, managing up to wade docs. So it gives like all this context in the back end that can help simulate people's feedback, again, to make sure they're sharper, clearer, and better at unlocking decisions. So I love this one because it's, again, just helps enable everyone that's happier. [31:37] effects within the organization. I think that's how you can become really an AI champion and transform your org is by starting with your own wins, you know, start with your own meeting prep or whatever, but then be like, okay, how can I enable the next set of folks on this? How can we make this like org wide? And so I love that. I love that this is really something anybody can use, you know, on the, I, I created it in ChattoBT for many reasons, but one of them being that I don't have context to the conversation. So folks can really feel comfortable putting all their information [32:07] and making sure that no one's on the back end. [32:10] reviewing it. I think we can see some analytics here on, you know, [32:16] 278 people have used this to sharpen a strategy doc. So again, it feeds back into things like [32:23] my impact reviews and showing that I'm enabling the whole org. So I love this one as well for just

[32:29] making sure our meetings are more efficient, making sure I'm not a bottleneck. [32:32] And then I can still provide my coaching where it makes sense. And I can still have those people where I'm thought partnering on with their docs. But this helps me scale me basically and scale the execs before it gets to a meeting. Yeah. And what I would say is people also love when you come to an exec meeting or a feedback session where you say, oh. [32:50] I've already... [32:52] checked it against our strategy or have already tried to do a loop of this with this gpt like just that extra effort to go through an independent loop before taking synchronous time to get feedback is both probably improving the quality of your work but also just saving people's time and people really appreciate it so i think that sort of initiative is also useful and what i want to go to for our last use case is really [33:19] You've extended strategic thinking through the organization with [33:25] another tool. So I'd love to see kind of our last strategic alignment tool that you've built using AI, because I think it's a really neat one. [33:34] Yeah, so one thing we just launched about a month ago is in Notebook LM. [33:41] So this is enabled through Google. And this is like the announcement we made in Slack to give you a high view of why we did this and what it is. [33:52] It's basically a strategy companion, which we know that folks have a hard time looking at this big picture strategy work sometimes and then saying, OK,

[34:03] How does that impact me? And sometimes it's just hard to find the answers you're looking for within all these different strategy docs, all hands, meetings. And so... [34:12] We gave folks this basically knowledge base. Here's a screenshot of it, of what these look like in reality. I just, you know, cleared out the summary of our strategy to make sure I'm not like, you know, totally revealing everything here. But what's great is that we can continue to add sources over time. So you can see we've got a few dozen sources in here, which are everything from the [34:36] top-level strategy doc to all hands we've had to transcripts from [34:41] other meetings we've had around strategy, to every org's strategic action plan. And so folks can go and interact with these. [34:48] but they can also interact with this in a chat capacity and ask it anything they want. So here's an example of it in real life. So, you know, I'm saying, [34:58] I just gave it a... [34:59] simple prompt of, as an executive assistant, how can I contribute to Zapier's 2026 strategy? [35:05] And it's saying, oh, that's a great question. I think you can help with champion clarity, focus, and speed. Make sure we're spending time on the right priorities. Make sure that you're driving internal AI transformation, [35:18] And so I love that it gives me some of that and connects back to the sources. But there's also fun things here like... [35:25] This, it auto generates a podcast. So I don't know if I'm fully sharing my computer sound here, but I'll play it for a couple of seconds. This is fully AI generated just based off of the back of these sources. So it talks like this.

[35:38] Welcome to the Deep Dive. Today we're really giving you the essential shortcut here. Absolute alignment on the strategy. We're pulling the core ideas straight from the Zapier 2026... [35:49] and all this is AI generated of things you can create. So it really helps [35:54] make the strategy interactive instead of a static doc and static thing and helps folks get their questions answered again before going to their leader or going to someone in their org. And so I love this for just enabling folks to [36:09] be able to connect their work and be able to query this over time. And it's something that we can keep updated and [36:14] and make sure that it has the most recent information so everyone can [36:17] can get value out of it. [36:19] So Courtney, I love these use cases. What I keep reflecting on is people are like, oh, EA's are going to go in this age of AI. And I'm like, have you worked with a fabulous EA? Because the second they automate one task, they figure out 10 more that are so high leverage for the organization. [36:39] culture carrying behaviors, strategic like communication, operational efficiency like [36:46] You are just demonstrating that this can happen at a next level. And so zooming back out, what we saw today was everything from helping yourself dig out of a busy calendar with meeting prep to enforcing your cultural goals and leadership norms through always on feedback. [37:05] I [37:06] checking feedback ahead of a synchronous meeting so you can make sure it's aligned with both how executives want to receive it, but also kind of the important business initiatives and goals of the company. And then finally, how you can take all this content that's always, I mean, I just don't know an organization that is not consistently writing strategy documents. Just one strategy document to the next, big strategy document, little strategy document, strategy

[37:36] creating a purpose-built repository for that information that can then be accessed in a multimodal way for people to learn, [37:45] align their work, be educated, all that kind of stuff, I think is... [37:50] Super awesome. So these are really, really great workflows. We're going to do quick lightning round and then we will [37:56] get you out of here. [37:58] I think the first thing is kind of this thing I said, which is a lot of people think your role is going away or they're afraid of AI. [38:07] taking over [38:09] this role. And I think you're showing actually, you're just becoming so much higher leverage, you have to be having so much more fun. So tell me, how do you respond to to that feedback around AI, and in particular, the kind of EA role? [38:23] I definitely am a believer that AI can only enable us in this role. I think it's a... [38:30] When, not an if. We will have to be folks that adopt these tools. There is simply too much to do. I think there's the one of the biggest problems that EAs have is we always have more work that we want to do than we have time for. And so that's what it's consistent. I'm consistently hearing. I don't have enough time. I don't have enough time. Like this is how we can do that. And [38:50] I mean, to I don't like talking about myself very much, but to humble brag and tie this back to real impact, I've. [38:58] I've gotten three promotions since I've been here on Popier and been able to work myself out of multiple roles and [39:06] you know, I think that this is how you can do that. You can be this really great AI partner for your company and, you know,

[39:15] Make sure that you're at the forefront and take all that busy, manual, repetitive work off your plate so you can do the human stuff, the fun stuff, the stuff you like to do, the stuff that's creative and – [39:25] relationship-driven and those things that we wish we had time for. And so [39:32] I think that it's just... [39:34] It's such an exciting thing for our role, and I don't think it's going to take our jobs. There may be, you know, certain... [39:42] admin things it does take eventually, but it's not going to take the whole thing. Think about the scope of what we do. Our whole job is to be, you know, all wide across the org and a little bit of depth in each area. But now you can be wide across the org, have more depth and have more time for projects and special things that you can do for your team. [40:00] Okay, so speaking of special things, [40:03] You have been one of the people I think is closest to answering this question, which is how... [40:08] close do you think we are to [40:11] replicating executives. How happy is Wade with your AI versions of him? And do you think we're actually working towards a world where those get quite high fidelity relative to [40:24] individuals, preferences, feedback, thought process, communication process? [40:29] Yeah, it's a great question. I think there are certain parts where we're getting kind of close on. So over time, [40:37] You know, I used to think of myself as a clone of Wade in certain instances. I could write comms like him. I could run his schedule. I could do things like him. And now we've got a clone of AI tools that are helping me do things like him, too. So there's certain parts of...

[40:52] of what he thinks and does that we're pretty close to replicating with these ai tools [40:58] What I'm not sure of is... [41:01] Wade or any exec, but speaking just for Wade here, he's a constant learner. He is... [41:08] He is ingesting so much information, learning new things, trying new things, building every day, like hands on keyboard, like the amount he can grow and learn, like the pace of that is so high that I wonder... [41:22] how you could keep [41:24] these models up to date with that because like I you know I know they they update fast but like how can you how can you grow at the pace of someone's brain and how [41:36] your evolving does change. Your thought process does change over time. So there's things, old things in Zapier history where someone's like, [41:43] oh, I heard Wade said, we'll never do X, Y, Z. And he's like, that was an old decision. That was an old decision. Like I have so much new context that's changing the way I think. And so for a [41:52] constantly adapting, changing the way they think, [41:55] and using new information to help them make smarter decisions. How do you how do you replicate that part of it? I don't know. There's things that but the consistent side, like how he writes internal memos or, you know, how he writes certain things are pretty close to replicating. OK, and then final question. [42:15] When AI Wade is not replying to you the way you want, or you're not getting what you need out of a specific prompt, what is your prompting technique? Are you an all caps girl? What do you do here? Oof, I'm not an all caps girl. I'm a ramble. So I love the dictate feature and I love to talk to it. And I give it feedback, very, very direct. I'm just very direct as a person and demanding. And so I do that. I'm like,

[42:42] I don't understand why you're doing it this way. Show me a reasoning of why you did it that way because I told you to do this and this is what I'd like to see and here's what feedback I have. And I just ramble and give it, [42:52] you know, everything I'm thinking, I think folks like sometimes... [42:55] don't know where to start and and try to give it this very specific feedback and write it no just like [43:01] give it top of mind pretend it's someone with no feelings and be demanding so i don't know hopefully the arrow ads don't come for me i do think thank you sometimes so good good good yes ai ai we uh we we know you have feelings i don't know we'll see what happens in 10 years yeah um okay courtney this has been so great where can we find you and how can we be helpful [43:31] about additional use cases, things I'm building, workshops we're having. [43:34] But I'd love for y'all to check out this EA ExecOps AI playbook I just made recently, which has like... [43:42] you know, six different categories of things that the EAs do consistently and different ways we've, we've automated that and ways that folks can replicate it and give me feedback on it. I'd love to, I'd love to hear what we're missing, what we can build out more. I'm looking for new things to build all the time. So I love, I love getting feedback from the community on what would be helpful. [44:00] Awesome. Well, thank you for joining us. Yeah. Thank you, Claire.

[44:12] You can also find this podcast on Apple Podcasts, Spotify, or your favorite podcast app. Please consider leaving us a rating and review, which will help others find the show. You can see all our episodes and learn more about the show at howiaipod.com. See you next time.