Using AI to Improve Productivity w/ Pete Christothoulou | builders podcast

0:00How do we reclaim half the work week for workers by removing the pain of all those mundane tasks and automating them away so they can go focus on your work stuff.

0:15Today on Builders is Pete Christothulu. He's the CEO of AI startup Zembli. We discuss productivity issues in the workplace, which tasks are appropriate for AI to solve, and how Pete's evolved as a CEO. Before that is my weekly check-in with editor-in-chief of the B newsletter, Marty Beckerman on the skills business leaders are looking for, a startup that regrows a body part, and what an organizational psychologist says about surveys.

0:46All right, once again, discussing three stories with editor-in-chief of the Bee newsletter, Marty Beckerman. Marty, how are you doing? I'm doing great, John. Great to be back. Nice. Great to see you. So the first thing that you're covering I'm excited about is this survey that was conducted of a thousand business professionals asking them what skills their employees needed to invest in or needed to have. Can you tell us about that? Sure. So managers are overwhelmingly saying that the skill they want employees to have the most is AI, knowing how to use it, knowing how to incorporate it in their day-to-day duties.

1:23The only problem is that half of employees say they don't know what they're doing with AI, don't understand it. that's according to a LinkedIn study. So there's a disconnect between the skill bosses are looking for and that employees have. But sooner or later, you know, the AI is just going to take all those employees jobs anyway. So why bother? Well, maybe, maybe not. I mean, what should, what should employees do if they want to bone up on, on AI? Should they go back to school or sort of what, what are their options? There are high level university programs. That's probably not necessary for most workers.

1:58Those are tens of thousands of dollars, but there are free and inexpensive resources for learning the basics from Google, from Amazon, LinkedIn Learning, and actually Harvard University has a course people can watch for free. And for only a couple hundred dollars, they can also get a certification if they pass a little test from Harvard, which would look great on their LinkedIn probably. Yeah, no question about that. Surveys. You cover surveys in your newsletter, too. How often when you receive a survey from your employer, Marty, do you answer it? Well, I always do everything my employer wants me to do, John.

2:39So I would say I fill out surveys a perfect amount. I'll need to send you a survey about how often you fill out surveys. Send me a survey about my survey experience. How was my survey experience? I will do that. Well, the interesting thing that I found out is in best places to work surveys, actually the percentage of employees that fill one out is a really big criteria or a correlative factor to whether or not it's a good workplace to actually be a part of. So if you're an employer and you want your folks to fill out surveys more often, what should you do? Right. Our organizational psychologist really dug into this this week, bringing up a bunch of studies and research that shows what makes employees more likely to comply with survey requests.

3:31The most important things are to notify employees in advance to let them know the survey is coming. This has shown to really boost compliance. make sure that the survey is easy to read, easy to understand, and even offering incentives, whether that's little gifts or free food and drink for completing it can also boost compliance a lot. All right. And then you cover a bunch of really interesting startups in the B newsletter. This one maybe takes the cake. It's a startup that if you lose your tooth, it envisions a future where you can get your tooth back without dentures. Maybe talk about this one and use the pun, please.

4:20The molar the merrier, yes. This is out of Kyoto University in Japan. It's a startup based out of the university, and they have demonstrated animals it is possible to regrow missing teeth. So the main purpose of this, some babies, their teeth never come in. This could help boost that so they don't need artificial teeth. But they actually have found that when an adult loses a tooth, they still have a little tiny bit in the gum and that this technology might be able to regrow that part. So whether you're a baby without their teeth or you just got knocked out in a bar fight, this is the startup for you.

5:04All right. Well, we'll be rooting for them. As always, you can check these out at the BE newsletter. And thanks, Marty. See you next week. Thanks a lot, John.

5:18I'm here with Pete Christothoulou, CEO of Zembli. Hey, Pete, how are you doing? John, so great to see you. Yeah, great to see you too. We talked last year on the podcast about AI and productivity. And since then, the company Zembley is rapidly growing. I want to talk about that. And we were joking off air a second ago that that was before when we recorded our first podcast was before OpenAI was launched. And of course, the discussion around AI and all the chatter about has exploded. And I want to talk about that too. to maybe reorient the listeners, if you would. Like, what's the pain that Zembly is focused on solving and what does the company do?

6:02Sure, so Zembly is an AI automation assistant. Really what that means is we've taken the power of a very seasoned executive assistant and chief of staff and wrapped it into one holistic product that can help workers improve how they operate and companies also improve how they grow. The problem that we're solving is, I think, impacts everybody. So a stat that I saw recently, which I still think is true, is that workers believe that 60 % of their work week is unproductive. So if you look at your calendar and you think about how you spend your time, 60 % of it is unproductive. And now they're also being asked to do more with less.

6:47And so, you know, that's a problem that you could understand likely gets worse over time. And the reason they're unproductive, honestly, is because the productivity tools on the market have failed them. Like all these kind of technologies that are out there, all these products that are point solutions to help you manage your calendar better, schedule a meeting or even take notes or, you know, track your tasks. Like all those things have kind of failed. and if you look at like what's going in the workplace the thing that we see is that work has become decentralized you no longer work in like one thing like most companies have somewhere between 150 to 250 apps that they work across those silos have created distractions for workers now you have bells going off in every different systems, Slack messages, Salesforce pings, you know, it just, it becomes like untenable for them.

7:47And what's happening underneath that is that even though these distractions are happening in these different systems and you don't want to pay attention to it, your team works across all these different systems so you can't get away from it. And in order to be successful during the week, you have to have a workflow that connects to those different systems, number one, but more importantly, follows you across all the disjointed tasks that you have. So if you think about the week, a day, a week in the life of a worker, it looks something like this. An hour a week on scheduling and calendar management complexity.

8:2930 minutes per meeting. At least, at least. 30 minutes per meeting to take notes or if you have a note taker that you use to review the notes and make sure that they were good. Most people are supposed to take their customer conversations and put that data into Salesforce or HubSpot. So now they have to make sure that that information gets there and that all the records are appropriately tagged in the right way so their team is up to date. The reality is most people don't even do that. They say, I don't have the time to put that information there. coming out of meetings about 50 percent of the work commitments or the action items that people say they're going to complete aren't ever complete they don't finish and probably john when you look back on your week you ask yourself like how can i make myself or that meeting more effective how can my teams spend their time better you know the reality is like probably half of your meetings might not be effective, but you don't understand where, how to improve them.

9:33And so really all those things I just articulated are disjointed tasks that everyone deals with. And so the reason modern tools have failed is because they don't have context across all those tasks or live in the different information systems. And so the problem we're really trying to solve is how do we reclaim half the work week for workers by removing all of those, the pain of all of those mundane tasks and automating them away so they can go focus on real work stuff. So let's give a couple of examples because, well, let me start here. You and I wanted to get together to talk. How did you actually schedule a meeting with me?

10:14Can you describe what you did? Sure. I just added Xena and I said, Xena's scheduled time to jump. And Xena is a conversational AI that goes and works with the attendee, if it's external in this case, to schedule a time. So I don't need to send a link, say, John, look at my calendar. I think that's a little impersonal. I can just add Xena like I would had I had a real-life EA. And it's handled perfectly in conversation. And it takes me about two seconds to ask Zena to do that. Yeah. So Zena sent me an email saying, you know, asking me or with Zena CC'd. I responded saying some times that were open.

11:01And then did you respond or did Zena respond to my email? No, Zena responds. I don't. At that point, I'm out of the conversation. I can see it. It's on email, but I'm not. I'm assuming it gets organized at the right time. so we schedule that's important is because my days you know consumed with other meetings different priorities and while i may have time that looks available i really don't want to meet at certain times and so the system has a real understanding of me and is able to layer in a meeting based on you know kind of my preferences of how i operate and so we we talked i i recall this meeting and um we talked about a few things one thing i was going to do one thing you you were going to do and um and zina actually sent me me a note saying schedule the time with pete about the podcast so uh so zina zina actually was my ea too for this particular um for this particular interaction it was it was really cool how how was zina kind of processing what happens in the meeting because the notes and the action items were very on point.

12:10Yeah, so first, we call this closed loop, which is, you know, you have meetings. The real goal in the meeting is not the conversation and back and forth. And John said this, then Pete said that, and they talked about, you know, pricing or whatever it is we talk about. Like that's not the interesting stuff. The interesting stuff is who committed to what and when does that thing do? And doing that with high precision consistently. So that's where we really specialize. And so going to the action item that you had, we really believed in closed loop workflows, which is people commit to something. those items were then distributed to the right owners and it's the product's job just as if i had a great chief of staff to go track it down and say john you said you're gonna go do this make sure you get it done is it done or not that's a really important part of the product and we see it really helps teams improve what i'll call goal completion by up to 50 because now the product's actually helping them kind of along the way the way we think about the product to your second to your question is really looking for what we call commitments to work.

13:33So what are the actionable commitments to work that were discussed during the meeting? Let me tell you something that is not a commitment to work. John, if you and I, you're coming to Los Angeles and we want to go do something, have a good dinner like oh yeah john there's this really good greek restaurant called luca like let's meet there on friday you're like oh yeah cool i'm gonna see you there friday well that is not a commitment to work it's an action item but that's not a commitment to work so we don't even include that in the back and forth of meeting notes because it's not relevant uh or in meetings you may say something like, yeah, well, okay, shit, I'll get you that on Tuesday.

14:24That's how people speak in meetings. Their English isn't proper. They don't dereference pronouns. So Zembley's job is to figure out, like, who is I? That's Pete. Who are you? That's John. What is that? That's the slide deck we talked about. When is Tuesday? Well, that's Tuesday, October 23rd. That's the product. So we really focused hard and intently on uncovering commitments to work with very high precision. It doesn't mean just getting the item right. It means not knowing what an action item is. So users then don't have 50 action items coming out of a meeting. And if we weren't good at that, the cascade of negative actions would be really bad.

15:13So now going back to the closed loop moment that we talked about, John, imagine we have a meeting and we have 10 action items, but six of them are irrelevant or not really the right things. Like we said it, but we really didn't mean it. Well, if that information is cascaded into Salesforce and now there are tasks for everybody and Xena is trying to follow up with you, that's at the point where it becomes annoying and technology loses its power and isn't helpful, actually. It's a nuisance. And so we work really hard about that core, what I'll call high precision, high specialization of the task in order to help workers.

15:59And so one more question about this. So I just had a meeting right before this podcast, a meeting with my team, and we were talking about strategy and having a discussion. And one of the things that occurred to me at the end of the meeting was like, what is what is the team think about the strategy? And and what did the team think about the delivery of the strategy? Were they, you know, were they inspired? Were they not inspired? Whatever. How does how does Xena help with that? Or is that something that you're kind of actively developing against? No, we do. There's two ways that we handle it. One, we can send out an anonymous feedback poll post-meeting.

16:43And you could have that set up for one meeting. You could have it recurring. You could customize it to when and where you want it to ask. But what you can ask, you know, send out an anonymous poll, and it's up to the users to give feedback against the variables that you articulated. The other thing that we do for managers is we give them a data-driven outlook to improve not just like one event, but all their events. And so I'll tell you what we don't believe in. We don't believe in giving you things like sentiment for meeting. How do people feel today? What's the mood of the meeting? like it gets a little squishy and honestly like not actionable you could you could look and feel like not amazing but you're having like a very critical conversation that's important to have and you're engaging differently so the things that we tend to focus on are what are the elements that we can automatically like pull out of the conversation to give the leader insight into whether or not that was a good event and if they should keep doing it or not so an example would be how many action items do we capture on average from that meeting?

17:57Is it one or is it seven? If it's one, is that really a good use of time? What's the cost of that meeting? How many people, how frequently, how many hours, what's the dollar amount? How many people speak during the meeting on average? Is the meeting shared or not? And other variables. And so we take all this data and we organize it into like a meeting effectiveness index. And we'll say something like, hey, John, here's the effectiveness index for your team's meetings, not just yours, over the last month. So there's a bunch of data. We're not going to tell you that a meeting was good or bad. We're just giving you some information so you can have an open dialogue with your team and say, hey, here's some information we see.

18:40Let's go deeper. Should we be doing this still, not doing this? How should we change how we operate? Yeah, I like using the metrics about what is said and the actual content of the meeting. I know that in some cases, I know that technology is kind of rating how much I make eye contact. But I look up and to the right when I think. And then sometimes when I'm deep in thought, I look constipated. Even though I'm actively thinking. So a robot kind of like interpreting my facial expressions, I don't know that I love, but I do love the idea of understanding engagement, understanding satisfaction through hard data, hard feedback that folks are giving me and what's actually said in the conversation.

19:31um you know discussion of of ai is is everywhere it's kind of like thinking back to our previous conversation i don't i don't think that we just said ai ai ai but now there are are and ads during football games where salesforce and other companies are ai is like the main message where do you feel like the media discussion or general discussion of ai is either misinformed folks are getting it wrong? How do you think about this excitement with respect to quote-unquote AI? Yeah, really good question. First, I think we can all believe that AI will help us in amazing ways and will do very powerful things.

20:19So I think the consensus, which I agree with, is that life can be better and things will improve for us. And I'm hopeful for that. The thing I think that's being missed, and if we're not careful, I think we can get in a lot of trouble is that AI will give rise to more silos, more data, more distractions, and will place even the greater demands on our time and attention. So, you know, I talked about this earlier about our problem, like the real practical problems that plague workers. and even more challenges with AI. So, you know, the world doesn't need an AI companion or co-pilot to summarize my Slack channels.

21:04Like, you know, or to gauge the mood of this conversation. Or if I join a meeting late to recap the first five minutes or to compose a message for chat that I want to send to you. Like those are like nice bells and whistles, like they're iterations, I'd say. And those iterations of like the products evolving actually can create more distraction for us. The world needs like a revolution that frees them of these tasks. And so, you know, I think if I was to, you know, we're all fans of history at some level. If I was to wind the clock back and I said, and I tell you, hey, John, you and I fortunately worked together in the 90s.

21:51Hey, John, the Internet's going to be amazing. Do you know, like, imagine that you have a pocket computer that holds all the world's information that allows you to connect with anyone, anywhere, anytime. Isn't that going to be amazing? And you would say, yes, we're going to be incredibly productive as a society. we'll be able to do more with less life's going to be good well that happened like we have those things i think the iphone gives you the power of a like the apollo 11 computer in your pocket now something like ridiculous like that what's also happened is we are so distracted we're busier than ever we're i think we're in uh something like the time that people meet is now double than what it was 20 years ago.

22:41We're more distracted than ever. You know, like, and as I talked about earlier, like we're working harder and we don't see as much for it. And so I think if we're not really careful, AI can take us further down that rat hole. And I think what you're going to see are companies, whether they're large or small, building AI bells and whistles to say they have something an AI and not building products that have are specialized are precise that work in the way that you want to work that are available on demand and actually autumn truly automate for you I think you're seeing like the first revolution of cool products like I can do cool stuff with it it's kind of like the first time you got on the internet you could see a store now they have an online store.

23:27Isn't this cool? And, you know, like, I think, I think it's going to take a year or two where you see like the, like the real companies start to separate and real power deliver. I'd love you to share if you can, what you're working on now at, at Zembly. I, I think it's tasks. I don't, I can't, I can't quite remember the word for it. And then also, are you taking on customers right now. Last time we chatted, you were kind of in a closed beta. Has that changed? What if someone hears this and wants to try Zembly? Yeah, they can go to Zembly and we are testing our self-serve. So we've sold to the enterprise predominantly with many companies that people have heard of, from Salesforce to Twilio and other big companies like that.

24:17We are testing self-serve, meaning allowing anyone to come in and say, I want to try it. so if you hit our site at the right time you may you may get the opportunity to do that and if you don't do it immediately like in the next couple weeks it'll be there permanently so yes you can go sign up you know the things that we're working on are really like further specialization of the product and so we built a compound product in other words it does many things we've not built a point solution and so the things that we're doing within the compound product are really trying to specialize against the practical skills that we've already created so if you think about it assembly has six or seven top level skills we can schedule meetings and manage your calendar and we can optimize your calendar and how you spend your time we can capture your meeting notes but importantly like the action items we can make sure that you manage your priorities or tasks.

25:20We call it smart task manager really well. We can provide coaching. All those things are top level skills. But within each of those, we have hundreds of ways to use the product. So scheduling a meeting isn't just, hey, Zena, schedule a meeting with John. We need to be able to handle lots of ways that people speak. We need to be able to handle non-continuous day ranges before or after this date or that day. We need to be able to handle little things like, schedule a meeting with John, but make Ziad optional and even handle like multi, one to many meetings. So for example, for you, John, you could say something like, Zena, schedule a one-on-one with my team weekly for 30 minutes.

26:07So one to many. And so we're going really, really deep in the skills that we have to uncover them as one. And the second, I'd say, is very deep on where Zembley is available. So we want to be kind of on-demand, available in the tools that you operate in, whether that's Notion, Coda, Jira, Slack, HubSpot, wherever it is, we need to be available there. And then finally, it's just closing the loop on the tasks. So once you ask Xena to do something, it's not only done for that task, but the cascade of things that need to happen to really kind of round out the experience in the same way if you had a real live BA or chief of staff, we can handle.

26:55So Pete, you mentioned we worked together in the 90s. I'm going to stipulate for the audience it was the late 90s. We're not that old. But I've known you for a long time and also known you as a CEO. And I wanted to ask you, how are you different as a CEO compared to, say, 10 years ago when we were actively working together? And as you're thinking about the answer, I'll paint a picture for the audience. I might have walked into Pete's office and he had an iPod. I think it was an iPod or something like blasting in excess. And Pete's got a t-shirt on right now if you're not watching on video. But you'd wear a tie.

27:40You dressed up. I look nicer than I do right now. um you you have artwork behind you you had you had artwork back then too um so i'm i'm ribbing you a little bit but how are you how are you different in general as a ceo than you were a decade ago yeah outside of appearance um i'm probably in better shape than i was actually than i was okay so i'm taking better care of myself that's good uh number one um but how am different. I think that's a, it's a great question. Um, so I, I've tried to have a growth mindset. Um, kind of is like the way I roll and to do that, you know, you have to be open to learning and being curious.

28:26I think that was harder for me when I was younger. I was a little bit more focused on my own opinion. And I would say strong opinion, strongly held. And I think, you know, the way I've evolved, one, I've had, I work with amazing people. Like it's, I've been really, really lucky. You know, from the team that I have here, who was you and I work with together, to you and many others. And I've gotten to learn and I think I've gone from, you know, from strong opinions strongly held to strong opinions loosely held. And I really, really tried to let the team guide a lot of what we do. I have, you know, as you know, as I think about product, you know, trying to combine what I'll call, you know, data, like very data driven, customer centric approach, overlay with instinct.

29:32And the instinct part is really important, especially at the early stage. And so what I try to do is bring a little bit more of that into the conversations that we have. so people understand like where I think the product needs to be. And look, I talk to 50 customers a week and I think that's something I didn't do before. I got a little bit consumed with what I call like the business of operating a business. And I'm, of course, consumed with that now, but in a different way. I'm very connected to, I think, what our customers are saying, how they feel, whether they're live or we're in their pipeline and just trying to bring that information back to the team so we're making good decisions.

30:20Yeah, that prompts a question for me, which is like how you make decisions. I mean, the way that I manage my calendar, manage my time or try to be productive, I suspect is very personal to me. And we all have these personal things we're trying to solve for, But you're you're you're I don't know if that's true or not. That's just that's just what what comes to mind. But you're trying to solve like a massive, a massive problem. And so how do you how do you think about with the limited resources you have, you know, making the right decision about how to add a skill or, you know, or whether your summary should be longer or shorter, those kinds of things?

31:01Yeah, there's something freeing in our product that I've never had before. And that is we hear from customers every day in their voice what they want. So whether it's in Slack and they're asking Xena to do something or it's an email and they're trying to execute, like, you know, connecting people. we have real-time feedback and we know what people want and what we can support or can't support so we have a nationally customer guided view of the things that we need to work on number two the way we connect with our customers the way that we support them is very it's real time so there's lots of technologies out there that now let you set up you know what I call like real-time chat groups with your customers whether it's discord or slap connect or other we have that type of real-time communication with our customers where we can field questions and also provide tips and see how people react to the tips so we you know I'd say is we have three or four different channels that have ongoing customer communication that we're able to take and like inform what we do the hard part is like the prioritization of it so you know i'd say it's most companies have a tough time figuring out like the sequence of what they should do i think we don't that we don't have that anymore like we have that nailed it becomes a little bit more challenging when you have to figure out like in what sequence is important because everything is important so that like as you know there's no easy way around that except a good old-fashioned team conversation to battle it out, which we still do.