Name: Openai On Openai How Our Sales Team Uses Chatgpt Work Final
Uploaded: 2026-08-07T18:37:58.017Z
Duration: 37 min 59 s
Description: Openai On Openai How Our Sales Team Uses Chatgpt Work Final

Transcript for "Openai On Openai How Our Sales Team Uses Chatgpt Work Final": Hi, everyone. I'm Connor, head of sales at OpenAI. I'm joined today by Erica, account director on our sales team. And we are so excited to be with you here today to tell you about how our sales team uses JWT work and the way that it has transformed our business. And as you know, the best sellers win because they deeply understand our customers. They understand what matters most and how to prioritize it, and they are always one step ahead of the next conversation. But this is very challenging because sellers have to do this across an entire book of business, and that even today can still require a ton of manual work. And it is this manual work that takes up so much of their time keeping them away from sellers. They're stitching together information from a variety of different sources. And similarly, leaders lack the signals and the insights that they need to run their business in real time and act before the moment to influence outcomes has passed. But at the same time, we are in an extraordinary moment for revenue teams. We are living in an era now where intelligence carries across every account, every workflow, and every decision. And this is changing the scale of what sales teams can see and understand and accomplish. And it's certainly changing the scale of their ambition. I know that's true for us. And this is something I get so passionate about because with ChatGPT work as the home that has been designed for us to meet this moment that is happening in revenue organizations today, we have a chance to change the way that sales teams operate. Now sellers have the ability to capture information and carry it forward across a variety of different sources. They're able to use all of this information and take it to be more strategic with customers and with cross functional partners. And then leaders also now have the visibility that they need, accessing real time information that they can use to strategize and run their business. Rather than having to rely on pipeline reviews and forecast calls that often become stale or static pretty much the moment that they leave them. And so within this world where you're using ChatJBT work, every seller now becomes a major source of insight for innovation, and leaders themselves are now redesigning entirely new workflows, agentic workflows, and how they run their business. And that's what we're here to talk about today, and we're seeing this firsthand in our own sales team and with organizations that we're supporting who are changing the way that they learn and the way that they work. And we're going to use a very real example that I think will resonate with any of you watching this webinar today. Setting the stage. It's a new quarter. You have a new book of business. Mhmm. You have a new number. It's a bigger number, and it's a target that is not waiting for you to catch up. But I don't wanna put words in your mouth. I'll let you tell it. Over to you. Yeah. First, I'm so excited to be here with you guys. I am thrilled to talk to you about Chatt GPT work. First, because it is truly my favorite thing and because it's changed the way I work in a fundamental way and really altered my perception of what the AD role can and will look like in the future. My husband's a software engineer, and I watched his job completely change in the last year. And when I first was exposed to ChatGPT work, I was like, oh, my turn. So let's take a look about what we're talking about here. So first, this might look familiar to you. This is ChatGPT. If this is familiar, you might be one of the roughly 1,000,000,000 weekly active users. I know I am, which means roughly one in eight living human beings on planet Earth will come to this interface in the next seven days asking for help with something. But what's new to help them is ChatGPT work. So we're talking about how transformative this is, and I wanna explain it in terms that sellers really understand. So we first gained access to this iteration of the technology in q one. And, admittedly, at first, I didn't use it. I was like, I have chat. I'm good to go. And then these new hires came in, and they were very chattyPT work build, and they were smoking me. They were smoking me in terms of quality of work. They were smoking me in terms of the amount of work that they could get done and the size of book they could manage. They were leaving the office at a reasonable time. And worst of all, they were smoking me in terms of revenue generation. And I didn't get into sales because I'm not competitive, and so I made a focus in q two to use ChatGPT work for as much of my job as possible. Now what did those results sound like in seller terms? I'm gonna tell you in terms of quota attainment. In q one, I used just ChatGPT, and I did a 138% to goal. In q two, I used ChatGPT and ChatGPT work, and I did 489% to goal. Congrats again. Thank you. Awesome. My entire team was heavy users. It is the most up to date my CRM has been in my entire career, and I have not been native in my CRM tool in two months. And my team, one of many that reports up to Connor, had a 97% forecast accuracy because of this. And it is my opinion that this technology is better for the AD, better for the business, and better for the customer. So what can take you from a 138 to 489%, give you forecast accuracy within 3% margin, and be better for ADs, the business, and the customer? Let's take a look. So this, as I mentioned, is ChatGVT Work. ChatGVT Work is your personal agent. It has access, secure access, to all of your enterprise data, access to the tools it needs, skills to use those tools accurately, and the ability to take on multi step tasks on your behalf at scale. Meaning, you can send this agent off to do 20 things at a time, and it can go help you with all of them. And I know Erica is gonna show this is new to many people. Maybe you haven't tried ChatGPT work yet. There might be new terms that we use, so that's why we have the chat here. That's why we have a q and a after. So ask questions in the chat. We'll have a q and a afterwards, and we'll make sure we get to those as we can. So the first problem I'm gonna show you that ChatGPT work helped me with, as Connor mentioned, is a new quarter. So what happens whenever you do a great job on your number? They raise your number. Sorry, but I am not sorry. That's what you get from being a top performer. So with that came a new role, yay me, but also a big change in my book of business. These are actual numbers, guys. This is OpenAI on OpenAI. Everything I'm showing you today is not customer data, obviously, real. So I started the quarter. I had a 127 accounts entering my book, a 105 leaving, 48 open ops entering, 58 open ops leaving. As you know as sellers, that is an incredible amount of work. So that is hundreds of accounts that I need to package transitions for, provide them to new ADs, provide intros, receive intros, receive transition information, and as you know, my number starts immediately. The the second that quarter starts, I'm expected to and expected myself to produce revenue. Lucky for me and now lucky for you guys, I have access to chat GPT work. So I went straight to my agent to get some help, and I said, again, this is my real prompt, identical prompt to what I used in real life. Okay? So no prompt judgment in the chat. I said, I need your help. Create a Google spreadsheet for me to map account transition. I have to stop you. I'm so sorry. How many of your prompts start with, I need your help, exclamation mark? Truly, like, 90%. I mean, it probably helps. I just had no judgment. I bet Chatty b t work is, like, chill out, girl. List I need a list of accounts going from me to a new AD and the reverse important fields would be account name, please link it to Salesforce, whether they're a customer or prospect, chat ARR, API ARR, open ops if any, next upcoming renewal date, when do I have to do something by, transition status, original AD, new AD. I said search my emails and find out if any of these account pass offs have already happened and if so, please link them. And take your time, make this document super thorough and organized by revenue as always. So then I did something that's really special to Chat two v two work that we couldn't do before, which is I provided a steer. And in my steer, I said, for all of my accounts, estimate the depth of relationship. Smart. Is there any sign that I've met with someone at this company? Search my Gmail. Search my, sorry. Search my email. Search my calendar. And take a look if there's any sign that, I've connected with them. And I'll add a little color here because this this is new and one of my favorite parts of ChatchaBT work. And we're not showing a live example, and some of this was not exactly live because you can see here, it worked for twenty two minutes. ChatChBT work is just that. It does work for you. You are delegating agents all over the place to do that work, so not all of it can be live. But a steer is what's happening when you have an existing prompt working. And you and it when I used to use, you know, chat, I would say, oh, I wish I forgot to add this one thing or and then I would have to wait for the prompt to finish. You can just steer it now. Feel free to just type it right in. You'll see the option, add to it. You are not fundamentally changing the prompt. It will still accomplish its goal. It just has this extra little steer, and it'll do that and incorporate it into the output. So make sure you do that. It's awesome. Totally. Something I learned from Peter Steinberger, our colleague and the creator of OpenCloth, is he has this really beautiful concept of having empathy for your agent. And the more that I think of ChatGBT work as a a coworker and helper, the better I am at working with it. And one example of that is STEERS, which is the same way if I asked a colleague to help me with something and then ten minutes later I thought of something, I might pop in the room and be like, by the way, can you see if I've actually contacted any of these accounts? Let's use that in when we're prioritizing transitions. So I kicked this off, before going into a customer meeting. It took twenty two minutes, and this is what it produced. So what you're seeing in AI land is what we call a one shot, which means ChatGPT work did this on the first try. This could be way more beautiful, but I really wanted to keep in for you guys a chance of what it did really on its first pass. So what information do we have? I have the account, and it links to Salesforce. If it's a customer or prospect, total ARR, breakdown of the ARR, total open ops, information on the open opportunities, transition status, so it says intro sent. It links directly to the email. So I have the intro email. I can respond from there. For accounts I'm passing off, it included relationship depth, developing, established. It even so you see it created all of these tabs. It even provided its methodology. Like, how is it defining depth? It says deep is if you've met with this person twice and you've had bidirectional emails, you know them. If you have literally never met with them and there's no sign that you've ever emailed, like, maybe you're not buddies just yet. But this is really cool if you can if and, I mean, we can see it here. It's something that I really like about ChatBG work is that it bakes a lot of the validation that you want from it into the output. So what I what I mean by that is, in the early days of this technology, we all had access to ChatBG, and let's be fair, we had to verify the output sometimes. Hallucinations were real. We wanted to make sure the information we were putting out there into our customers was accurate. And this is something I so rarely think about anymore because when you build something like this, it's like, oh, did it actually capture my accounts? It did. Salesforce links are right here. Or, oh, is it sending the right emails? You went in there. It's the right recipient. So it's doing these, like, confirmations for you in the moment that let you feel really comfortable about just letting this thing cook. It's really cool. Cool. Totally. It's it's hard to describe how helpful this particular use case was for me. And let's review what Chat GPT work was able to do in twenty two minutes while I was talking to our customers. It went through Salesforce, found every account leaving and entering my book. It identified their opportunities. It identified their ARR. It provided their usage information from our usage dashboard. It went through my emails and identified which ones I've gotten pass off emails for. So cool. It provided links to all of the emails. This is pretty pretty impactful stuff. So let's say we've done a great job. We've done all of the account pass offs, and now we're cooking and we're just working with our book of business. I have so many emails coming in all the time as I'm sure we all do. And so, again, I'm going to turn to my trusty assistant in ChatGPT work and say, this is a a plug in. We'll show you more later. Help me prioritize my emails. Who should I reply to ASAP? And the really nice thing about our ChatGPT work agent is that it doesn't even mind when you have typos. Yes. I it made me maybe a worse speller, but I feel like these are the trade offs. So you just need to be willing to accept. I don't know. So what you're seeing right now is an example of agentic delegation. When we look at what happened to coders, so none of us know transparently what our jobs are gonna look like in one to three years. But the best guess any of us have is taking a look what happened to software engineers and coders, which is it was kind of a a small adoption and then once they hit agentic delegation, once you had the ability to delegate multiple tasks to an agent, it was a complete role augmentation in a matter of months. That is what's happening now for all of knowledge work. And what you're seeing is an example of that. I'm delegating a task to my agent, and if we weren't doing a demo, maybe I'd fire off five at a time and then hop on a customer call. So let's see how it did. It went ahead and pulled three emails and just recapped them for me. So what's Victor's ask, what's the critical action, and I'm gonna get some additional help. So I'm gonna say, great. Draft replies to all three using my Erica's voice skill. We'll talk about what this is. Don't worry. And have all three drafts ready for my review in my email. And you're just at mentioning that. I'm just going at and then clicking Gmail. Yeah. Exactly. So let me see what these were. Alright. So I'm gonna say tell Victor tomorrow works for me. Tell Dana I'll check-in with legal. And for Rachel, I pinged Luke from IT about this. And thank Dana. Day. If you're watching, Erica will check-in with legal. She is very persistent internally and externally. Don't worry. That's gonna be a good problem. I've often described our role as your internal cat herder Yeah. And I will herd the cats. And for this one, I'm gonna tag in Slack. One thing I want to note is that for this example, I'm manually tagging all these in. ChatGPT work knows when to call what. You won't necessarily have to do this, but I want you to fully understand all of the different interfaces that it's working from and how in q two, which was a big result for me in my opinion partly because of this, I was able to live out of one centralized location. So while this is working, I promise I'd tell you what these are. Let's talk about it. You saw me use the Erica's voice skill. Again, like Peter said, have empathy for your agent. You have to teach it skills. You can't expect it to know how to write like you. I think of it as, like, if you just hired a a copywriter and you never gave them sample content, they're gonna sound like themselves, and they're gonna sound like themselves for everyone, and that's how chat is too. So I wanted to create a skill. It's the first skill I ever made for email voice. And I said, create a skill to make emails sound more like me. Go through my last 100 emails. Capture my voice, and anytime you're writing on my behalf, use this skill. That's so funny. I did the exact same thing. But I had it also I had it review docs I'd written. I had it look at all of the Slacks that I sent in all of the time that I've been here. But then it gives you a summary of how you sound to others, and that was revealing. It did it did the same for you. It did. And this is funny. In our IC channel, I'll let you peek in sometime. We all shared how it described our email voice. Oh, nice. And one of the descriptors for mine was sound busy, but in a helpful way. I I mean, that tracks. You are busy and you are pretty helpful. I felt very seen. I was like, I'm trying to be helpful, but so busy. That's right. So here is a skill. So now I taught chat g b t work a skill. Pretty cool. Learn skills faster than I do. In addition, we have plug ins. Plug ins are access to a tool and the skills needed to use it. So, if you meet someone and they've never touched a computer in their lives and you give them access to Excel, they're going to be pretty useless. But if you give them access to Excel and pre download every skill they need to be an Excel wizard, now you're really cooking. So we have hundreds of skills that are ready for you. We've made this so easy. All of these are ready to go and prebaked, but you can also do a group of of plugins as its own kind of entity. So we built one for you. Our team is really sharp and great at this stuff. They built a sales plug in. It's a myriad of tools. All of the skills that ChatGPT work needs to use them effectively, and most of all, the skill of knowing what to pull when. Yeah. So if I say I Slacked Luke yesterday, yesterday, it'll know to pull up Slack. If I say, did they ever reply to my email, it'll know to pull up whatever your email provider is. So this is ready for you guys right now. You'll just go to your side panel. You'll click plugins. Sales should pop up for you. If not, you can search for it, and you can, go ahead and download right there. So let's take a look at how, it's doing. It created some drafts for us, and we'll take a look in our drafts folder. Look at that. Three drafts are ready to go. And look at these time stamps. You guys can fact check me. We're live on the call with you at 09:46. All of these drafts, I can go in and edit and tweak as needed and then send. Alright? So we have now, gotten a good bearing on our book of business. And as is often the case, here comes a curve ball, which is a few days into the quarter. I shouldn't be surprised. I find out You do them every quarter. I should know that I have to do a QBR. So of my 50 customer accounts, 48 of them are new. And so it's a challenge to say, okay. Who are your top accounts? Where's revenue coming from this quarter? Whenever you have a brand new book of business. Maybe that was challenging the old way, less so today. I went straight to Chat GPT work. I, said, look at the information you have, and this is a a sales QBR that we built. Who should I select as my top five accounts? And it proposed a a top five list for me. Prior to doing this exercise, I had come up with who I thought were my top five, and four of five matched. And, admittedly, the fifth choice was better because my fifth choice was like, I'm gonna find the fifth one. And it it told me why. So 17,000,000 in in recurring revenue, 5,000,000 in open ops. These are the most important ones to prioritize. I said, great. Put these into a QBR deck. So as you saw, this is the template I was given, and it went ahead and pulled information for me. Awesome. It's remarkable, guys, how much manual work we have reps doing that are that's not necessary anymore. Necessary anymore. And one example is this kind of, like, data collection. So in this case, it went and pulled revenue data. It went into our usage dashboard and pulled usage data. It looked at the usage data and provided insights. It said this company is using codex a lot. There's areas of opportunity here. It went and looked at our our CRM data, LinkedIn, the company's website, and identified key executives. These are linked and created a best next action. Am I gonna agree with all of these? No. Sometimes I'm gonna tweak it, but it's really a helpful starting place. I knew something had changed when all of these QBRs were sent to me in advance. I it's like, I know I do not begrudge reps for doing these the night before you're all out in the field talking to customers, but that that was an AGI moment. I was like, wow. Okay. Something has changed here. That's what I think you feel that way about our CRM being up to date too. Definitely. It's changed the way we run. I I really can't couldn't be more excited for you guys to try it. So, I said this all looks great. Now I need an account plan as a site for each company. I'll talk about sites when I in just a moment and show you a little bit about what those are. So it went ahead and added account plan sites to each of my accounts. Oh, wow. Just like that. It was super fast. There's no going back for me. So here is the account plan. Again, guys, this is an exact replica of my real, work, and this was another one shot. So we have a CEO narrative. What are we talking to in terms of how we're selling to this organization? It did use case priorities for me. It went ahead and pulled, customer pain points. All of this information is coming from transcripts that we have. So what is everything that's been mentioned on the call? And finally, for the love of God, it is creating executive, org charts for us. This is stuff that reps used to spend so long on, like creating little boxes in in a Google Sheet. This is remarkable. This was definitely my favorite moment of the QBR when being able to review these. And I actually used this recently because I was going into a customer conversation. I knew we had these account plans, and I just pulled out my phone and just whipped this out on mobile because you can access all of this on mobile and just pulled up one of these account plans and was able to walk into that customer conversation and just, like, I was blown away by how much access to info I had. And then similarly, in the QBR, this was all sent in advance, like I said. So rather than me going into a QBR, and you never wanna do this as a sales leader, it's just asking me this question, update me on this, and just have you tell me information you already know. We've had it all. And so we got to spend that entire time just talking about how do we drive this forward? How do we find new use cases and help customers achieve these different outcomes? It's really cool. I love that. Totally. This is an exciting time because, you have an encyclopedic knowledge of every customer interaction. Every transcript is at your disposal. All of the notes from any rep they've ever worked with, all of the CRM data, everything is in one location. Like you said, you can use it on your phone, our usage data. So I was in the elevator on the way to an on-site the other day, and I was able to ask, what are their weekly active users like? Is it going up or down and get real time access to all of our information? So this, it built one for each of these accounts, and I think it did a really beautiful job. I have many areas on your team all the way reporting up. How are you getting your bearing on that scale? Good question. I guess, good segue. I I mean, I don't want done. Let's not diminish the work. You had you all have an incredibly difficult job. I've done that job. Congrats again. But similar challenges. We in sales leadership, we turn the page on a quarter and a half, and we are shaped with new environments as well. And it's across the business. And, of course, as sales leaders, we have to ask, okay. What's the number? What are we trying to hit? What's the plan to get there? Where are the risks? Where are the exposure areas? And, historically, that has been a time consuming task. You have to collect input and information from across the field, from your customers, from across cross functional partners, from the data. And it's not because sales leaders are not very good at setting strategy. It's not because sales leaders don't have confidence in their path to hit the number, but this can be a very difficult thing to do. And even if you get it all, there's a question of do you trust all that data that it's come, and do you trust that it's up to date and current? And this all changed for me with ChatGPT work and allowed me to sort of live out of that environment while I did all of this planning, strategy, and execution right from ChatchBT work. And what's so cool about this is oh, let me just back up and say, I didn't take my own advice that I give to folks who are new at ChatchBT work where I tell them, hey. Just start with a small prompt. Start with the workflow. See if you can iterate on it. What I did instead was, I did it my own personal AGI test, which is if chat or AI can do this for me, I know we are in a new way of working. I know we are in a new paradigm. And so I came and built a prompt. I'm gonna do voice prompts because that's what I like to do to save time. And I said, help. No. I I didn't really I didn't really do that, but I'm gonna start. But I said, chat chat gbt work. Please build for me a revenue command center. I want it to include, our current, revenue and I CAR. I want it to capture, week over week and quarter over quarter metrics. Please include forecast, build a talk track around that forecast, capture all of the information that you can from the field and develop a section that captures, product signals and anything else that a senior rev ops, leader and sales leader would want to have, available in their command center. And so I'm gonna stop it there. It's gonna write the prompt for me. But in the interest of time, I'm gonna jump right into the command center. There it goes. It built it. I'm gonna put it to work. And here it is. And so I'm gonna go to the top here, and here's the command center that I work out of every single day. This is where I start my day. This is where I end my day. And Erica talked a little bit about one shots. This wasn't a one shot for me. I worked on this a little iteratively, but it didn't take that long. And like any revenue leader, I come in, I join, and I wanna understand where am I in my business. How are we pacing towards forecast? But not just that. That's information you can get anywhere. I wanna understand what I can do about it. How is this changing week over week, quarter over quarter? How does our pipeline coverage compare to where we were at last time in the quarter? And I can see how these things are evolving in real time, and I can come in here. I have a handy little exec brief that I asked it to include because I'm not the only one that wants to understand my business. I share this information with executives and cross functional partners and just ask it to create a nice little tweet for me, a little TLDR that I can send out to folks. I can even generate it right in Slack if I want to. Similarly, it's given me forecast. And, again, this isn't just an area where I can see where I submit my forecast in our CRM. I wanted it to be dynamic. And so the cool thing about ChatGPT work is as sales leaders, we've used dozens and dozens of technology and tools over the years, much of which we love. There are all of us who have our favorite tools of choice. I just built the ones that I really like into this command center, which is just so still so crazy to me that I can do. And this way, I can do something we all do in sales leadership. We model out scenarios. Because not everything is set in stone. Deals will slip in. Deals will slip out. Not that that's ever happened to you. And we can say, what if our plan was bigger? What if we only realized so much of the deals that we have in commit, but we convert more of our best case or more of our upside? And I can start to just work with this dynamically and model out scenarios just to understand and be able to present a clear picture of the business when I go into my forecast. But that's not all. I also generate the actual walk up because we, have a very consistent way that we speak about our business when we go into forecast meetings, and I wanted to just be able to generate that consistently so I don't have to rewrite my talk track each time and make sure that I'm very polished when I show up for these conversations. I can just generate it. It gives the headline, what's moved, what's risk, and it creates it for me. And I think something that I would be remiss not to show because it wouldn't be a command center if it didn't have our top deals and commits, is just that. It's It's a place where I can review all of the top deals. And this one I spent a little bit of time iterating on because I wanted it to include all of the things that, you know, we all wanna know in sales. Not only who is the owner, what's the next step, how long has it been in a given stage, but the really juicy stuff. What are the use cases? What are the outcomes that we're driving for the customer? How are they engaging with our product? And so, again, CHPT work is, I set agents to delegate all of this work to collect this information from different data sources, product signals, rep, performance into one place. But I'm sure if you're watching this, you're also thinking, how does it does it stay up to date? Is this something that you have to refresh every day? So what I do is I go into, ChatchBT work. I'm just gonna make sure it knows what I'm talking about when I capture this. Let's say and say, update this command center daily at 6AM and 4PM. And so what I'm doing is I'm creating an automation. And, that automation is just just that. It's automated work that you can set at a recurring basis. It'll show up in my scheduled site as soon as it's completed. You can see refresh revenue leadership command center, and it's as easy as that. And I can do that with, regular tasks, and you can do that with sites. And as a reminder, sites are just that. These are web pages that you can build using Chatupi Chee work. They're dynamic, interactive, and they can pull from a variety of different sources, and you can use them for anything. Command centers, you can use them for account plans. You can even use them to replace decks, which is a cool way to do it if you so choose. Maybe the last thing that I will show here, I could geek out on this command center for another hour, and I have, but there are two more things. One is just you can see where performance is diverging. But, of course, all of this information is helpful, but at the end of the day, it's not going to help unless I can take actions from it. And so I can select the actions that I wanna take based on recommendations. I can adjust them, and then I can even generate the Slack once I've selected them, and it will Slack the owners of these decisions or involve the cross functional stakeholders and post it there for me. And then finally, I have my product signals. And so the most valuable thing that I have found or one of the most valuable things from ChatGPT work is it empowers us to be able to capture the information from the field, be better partners to our customers and to our cross functional partners, and give them the information that they need. And so with that, I know we're almost at time. I don't wanna get kicked out of the studio before we answer at least a few questions. The the takeaways that I would say are very similar to what I myself experienced at Attach PT for work when I started to build, with it and alongside it. I was able to see revenue risk and opportunities sooner. We're able to give sellers more time to spend with their customers. All of that manual work that is still so important can be delegated, and then we can keep the whole revenue team aligned because we have found a new way of working by designing new workflows, which HBT work at the helm. And so with that, I think we can go into q and a. Cool. A lot of questions on connectors. What are the back end connectors to enable this? We have 20 different systems storing customer info. We have over a thousand oh, and so, I'll take this one if you don't mind. Yeah. We we have, a thousand plus plugins, that connect to core data sources. So that's Salesforce, HubSpot, data warehouses, sales tools, productivity tools, and you can also build your own NCP connectors to connect internal data, that can be accessed by your employees. And so reminder, if you're a business customer, we are not training on your business data. If there are more questions, I'm just Connor at OpenAI. Ping me. If I didn't answer this correctly, I will get you an answer to that question. What else? Why would sellers use Chatchpad work instead of the likes of other coworker, AI tools? I'll let you take this one. Yeah. For me, it's functionality. It's remarkable how high performing this tool is. Whenever you're looking at a a lot of, other tools, some of them might be native in these different interfaces, and I think it's kind of bandaging over a wound. So what we're trying to do here is give you one epicenter where you can do all of your work from in a singular location. We know how to build things that are helpful. No one would be able to have approximately 1,000,000,000 weekly active users if it wasn't genuinely useful to people. And so I would try it out and see if it's helpful to you. And let's be fair. Many of our customers use a multivendor strategy. Like, you wanna reduce risk, you wanna complement model strengths and weaknesses, by all means, I think you should do that. I think reasons that customers use us more, especially these days is Frontier models. Our Frontier models are two to three times more effective per completed task, and we're very aware of how much cost efficiency matters right now for the customers that we're working with. That is something we're hearing a lot these days around cost efficiency, and that becomes the reason I think folks are turning to Chattopty for Work beyond the, reasons that Erica mentioned. That one example took twenty two minutes. Where can I see how many credits were used? Good news. We have usage analytics and controls. If we had more time, I'd go into the panel. But, again, I gave you my personal email or at work. You send it to me. I'm happy to show you. But we are typically, I I think as of the last look at data, two to three times more cost effective for the same task or outcome. I like the idea of it's taking twenty two minutes to do the work instead of spending that time is questions from Steven. Do you think about how not just to sit and wait for it to finish its task? Do you have a framework for which task and what order to do those tasks in? This is a very good question because when I first started doing this, I would just stare and watch the task because I was so excited, and I do not know that that was more effective. But it does it does change how you operate very quickly. And what I'm seeing with our sales team is they're starting to complete multiple projects at the same time. For my part, Steven, I now have a list that I use ChatouchBT work to create, and then I just start to delegate that list as I go down. And it is the most gratifying thing to literally scratch off the list after five minutes because all of that work is just underway. It's the same with how devs will work on multiple PRs at once. Yeah. I like doing them right before a customer meeting personally, so I'll have my list and I'll shoot off, like, five or six agent tasks, which is exactly the one you're talking about. This is exactly how I did it because there's something magical about leaving a customer call and coming back and seeing all that work done for you. But I'll tell you, even if you sit and watch it for twenty two minutes, that would have taken me, like, five hours to do. Yeah. So it'll still save you time. Andrew, are these sites internal, only available to team members, and private, is there a risk of pub, public access? All sites start only visible to the creator. So when I created my command center, I was the only one that could see it, and thank goodness because I did do some work on it. But they can be expanded, to your desired level of visibility. And IT admins can control this based on the settings within a workspace. Geraldine, can we have Connor's prompts for the command center? Of course. Yes. We'll follow-up with the resource. Happy to send it. Steve, how do I know if I'm using the right model for the job? Steve, that's a great question. I'd be curious to hear your answer to this because I know you played around a lot with this. When you think about it, we have Terra, we have Sol, we have Luna as far as the 5.6 series. And I I often think about it as, like, what is the body of work that I want done, and how quickly do I need it done? And so, for example, within that command center, it doesn't have all of the information. Sometimes they just want a quick, hey. What are the metrics for this week compared to where we were at last quarter? And it might not have the fidelity that I want in the command center. I don't need to go all the way to Ultra for that. I can use something pretty quick, maybe like Luna for that for that matter. Because I also want it quick. I'm walking into a meeting. I wanna make sure I have it top of mind. So if it's a simple question, go for one of the smaller models that are still very effective and more cost efficient. But if you're building the entire command center Yeah. It's okay sometimes to start with a at least a a one of the bigger, frontier models and make sure that you get the right work done. Totally. And when you're first getting started, when I was first learning, I would shoot off the same prompt with multiple models and just start to experiment with the results. And the more that you get used to it, the more you'll start to have your own opinion of I really like using a a more souped up model for this versus that. Yeah. Another question. Sometimes it's not obvious whether to choose chat or work. What advice do you have on this? Great question. You are not alone. When we first launched this, I wondered that. What I have come to as a realization is for quick chats where I just want a brainstorm partner, I go to chat. When it comes to a lot of what we talked about today, you're doing agentic delegation, you're creating spreadsheets, things that you know would take at least fifteen to thirty minutes, use work. But sometimes both can get a task done. And if you have time or if you feel like experimenting, give the task to both, and you start to get a sense really quickly of which one you wanna use. What do I prompt it to create something like this? Oh, just something like that you've seen. I would use chat to build a skill, and I I regret that we didn't get to spend more time with that because it's one of my favorite things. Building a skill is just that. You do it right in the prompt. You can even use voice, and if I was using voice and just clicking on make a prompt right now, I would say something like, chat to be cheaper work. You don't have to call it that. You can just say, hey. Please build me this skill. This is a skill that I want to pull in sources across my CRM and across my email. And every time I invoke this skill, I want you to also use my voice. And so if you don't know what my voice sounds like, please also review my emails and make sure that you captured it, and then send it to me before you publish this skill because I'd like to verify it. Models are really getting good enough where you can just start to brain dump. So please, by all means, brain dump. What is the difference between ChatchPG work and codecs? Is work just a non developer version? Yes. I don't wanna oversimplify or or, you know, be reductive in the answer, but that's right. Codecs is a is a phenomenal tool, for developers, and ChatouchBT for work is just that, but it's for knowledge workers. Same harness. And that's all our questions. I think we're getting kicked out of the studio. Thank you, everyone. And I think if I can leave you with one thing, I know that this can all be really new, especially if you're in sales. Sales are relate we we have relationships. We're very good at what we do. We spend our time with customers, and new ways of work can be a little daunting. Just jump in. Try it out. I promise it's exciting, and we're here if you need help. Thanks, everyone, for attending. We'll see you next time.