ChatGPT Workspace Agents Change Everything

So ChatGPT just launched its agents for work. So it's called Workspace Agents inside of ChatGPT. So they're introducing Workspace Agents. Teams can create shared agents that handle complex tasks and long-running workflows, all while operating within the permissions and controls set by the organization. This is all powered within ChatGPT. So you can see the different workflows here. Software review agent that triages software requests, enforces policy, routes approvals, and opens IT requests. Okay, that's cool. Third-party risk management agent that screens vendors for sanctions, financial, and reputational risks. And a lead agent that qualifies inbound leads, drafts tailored follow-ups, and updates the CRM. And a weekly metrics reporting agent. So not bad. I mean, this is, this is pretty cool. So I think if you're, you're looking to build this now, anybody can get this done and you can also integrate it into Slack. So this is similar to what we use inside of our Slack, which I can talk about later, but you can put agents to work across tools and teams. So you can see this guy, Zach Williams, what's the latest feedback on our mobile app? Here are the top themes from past week across public review sites, support tickets, and Slack customer channels. And so I'd imagine they're gonna probably add other integrations into this in the future. So you might have Mixpanel data, you might have other product data that you're pulling from different tools that you're paying for right now. And that's the value of this. And so they have these templates already, which I think is pretty cool as well. So they have Spark for qualifying leads, you have Slate. And so this team directory, it seems like you can share with people, which I think this is pretty Pretty cool. And you also have this analytics piece over here. So you have different— the safeguarding piece, the government, the governance, the permissioning, that part's going to be very important. So you can also see how active people are. So Victoria just completed a run. Ethan ran something here and you can see what's happening and then the amount of runs and the amount of unique users and how it's trending over time. So I think that's very useful. And earlier what I just mentioned, enterprise governance and visibility. So they come with enterprise-grade monitoring controls. Yeah. So admins can protect sensitive data while giving teams a safe way to move faster with AI. ChatGPT Enterprise and EDU admins can control which connected tools and actions each group can access or user groups can access. So anyway, I think it's cool that they're giving this away for free right now to give you the ability to trial this, to get you hooked on it. And then it's going to be on credit-based starting on May 6th. And so it's going to be basically usage-based. Okay. And so I think this is a good start and I'm going to show you an example of a chief of staff agent that I just built right now. And then I'm going to talk about the alternatives that are out there because this is, again, this is something that we've been using internally and we've been doing this for, for, I would say a couple months or so now with the agents that we have. And you can see the pricing here, how it looks. So you're going to have to get on some type of business plan, $20 annual or $25 per month. And then this is the chief of staff agent that I have over here. So literally what I did was I went over to ChatGPT and I have these agents over here. Okay. So you can see there's this chief of staff agent. So they have these, these templates over here. So let's go through them. For a moment and maybe we click on these templates over here. So for example, the chief of staff one that I just built. So prepare a high signal operating brief from schedule, inbox, and team chat context. Great for users who want sharper priorities, meeting prep, to-do capture, source links, follow-up guidance, and requested email or chat follow-up through one concise daily artifact. So this is what it works with over here. Fairly basic. So you can add Gmail, Google Calendar, and Slack, and then you can prepare an onboard— Operating brief from schedule, inbox, and team chat. Format a scan-friendly brief with to-dos and source links. And so I can hit use template. I've made this one already, so I'm going to choose another one to build. And while this one builds, then I'm going to move over to the one that we have and we can actually go through it because I actually haven't done that yet. So let's say I want to have a— let's see this marketing strategy one. So turn marketing asks into working briefs, stronger asks, clearer reviews, and better informed decisions. Create teams that need strategy, content, signal synthesis in one workflow. Not really what I'm looking for. Maybe I want something around knowledge surf. So search find answers faster across all different things that you have over here. Not sure I really need that one either. And these are fairly basic agents that you can build. So this one, maybe this one over here, sales assistant. Use generalized sales workflows for account intelligence, competitive research, value engineering, meeting prep, follow-up, pipeline planning, and seller coaching. Great for teams who want stronger prep, clearer strategy, and better execution. Okay, let's just do this one for fun and see what happens here. So we have Google Calendar, we've got Slack already, call intelligence. We could custom MCP into Gong, but I'm going to leave that alone for right now. CRM, we got HubSpot. So I'm going to make sure that that is connected. Okay. So just so you know, I got HubSpot. HubSpot is connected and let's just, I'm going to go back over here. So we got Google Calendar, Slack, HubSpot, Google, and for HubSpot, I had it all to view only and Google Drive is connected as is Gmail. And I'm generally okay. I mean, when it comes to security, OpenAI, you would bet that they've invested quite a bit in their just compliance in general, and they get a lot of those asks from their customers. And so I'm going to go ahead and set this up and go from there. And you can see the agent is working now, right? So start from sales template. I've created it. I guess we can skip to builder if we want, but I'm just going to let it work. You can see it's thinking over here. So while this is working, I guess it's working already. So you can see it's going through the initial instructions, the role itself. All these things, and I'm just going to have it at the base level, and I can use this within Slack. So that's kind of cool. Maybe I'll plug it into one of our channels later and see how it goes. But as it does this, I'm going to let it work. I'm going to move over to this one over here. So I did actually have it make a a document for the chief of staff agent. So basically, you can see over here it helps users prepare for the day. So I get full daily briefs and. Also just things to help me plan out my day basically. Right. So I'm going to go look for it and let's see what it looks like. So check this out. I actually have the brief. It took about 11 minutes for it to come up with the brief. And so as of April 22nd, the strongest end of day focus, here's what happened. Here's a few open loops over here. Here's what's happening with sales, for example, here, a couple of opportunities. Here's what we need to prioritize from a product standpoint. ClickFlow onboarding. There's maybe some speaking things that I need to get back to. And then also some things that we've activated from a just Agent fleet-wide, we've activated this new skill called Skillify and a couple other things that we're talking about over here that are popping up in Slack channels kind of across the board. There's also some recruiting stuff that pops up as well. But overall, and you can see what the blockers, the risks are. So this is actually, this is my first time seeing this and I would say this is fairly useful in my mind. And so I would probably add this to my Slack channel to get this every day. Maybe we have something in the morning, maybe we have something at the end to summarize. And so I can keep my priorities in check. And so I actually think this is something that I would continue to use. And so I think the key thing in my mind is when you, when you think about these, these agents, so the way we use these agents internally, we have our agent fleet is called Single Brain. And ours are revenue agents, right? So the institutional knowledge that we have around revenue, sales, and marketing, for example, is going to be up there. And so as long as you're constantly ahead, if you're building stuff like this, you shouldn't really be worried about OpenAI coming to eat your lunch because they're, they're going for a wider market, right? And so for us. One, we're gonna be working with our clients a lot. There's a lot of data that we're gonna have. And, you know, from a connector standpoint, they're just connecting at a, at a high level to the other software that you have, right? And so when we see our team members communicate with these other agents inside of Slack and they're able to do these data pulls, they can strategize with them and these agents can execute end to end. That's been extremely valuable for us. And that's where we, that's where we've seen the future of work going. People are like, we, we can't live without this, right? So I don't see us living within Slack, for example. I think it's, it's, you can put together or inside of ChatGPT, but you put these agents together and you, you have 'em live. Live inside your— live inside of Slack. And so the way Slack's or OpenAI is going to be making money off this is the, you know, the per-seat model and also charging on usage. And so they're going to get a lot of usage out of this, right? And so, you know, kind of similar for us. I mean, the way we see this is that we're doing a way— we're basically managing people's agent fleets. And so I think there's going to be versions of this where it's self-serve and you're going to have managed versions as well. It's no different than, you know, SaaS in the past, right? You're going to have self-serve versions and you're going to have versions where they need a customer success person, or you can even look at the for-deployed engineering model that Palantir has. And so this is, it's pretty cool, I would say. And so the sales assistant one looks like it can help me with a bunch of things, creating assets, drafting initial outreach, pipeline review, all these things over here. I'm just going to let it go. And then you can, I think the key thing to also understand is that you can continue to edit these over time. And as you work with it, it's going to continue to get better. And so for right now, I'm going to create, and then I'm going to go ahead and, you know, mess around with it and maybe I'll just connect that later. But all you do after it's gone through this is you hit create and you can choose if you want to schedule it or not and go from there. So what I'll do now is I can choose to run this if I want. So I can choose to let it create a sales asset, build from context, turn notes into follow-up, or mock up a workflow. And so I might just ask it, What's the most helpful thing you can do based on the context you have around us and the connectors that you have? So highest value thing I can do right now is sell our control tower brief for today, Wednesday, April 2020, April 2020, April 22nd, 2026. What still needs attention today? What matters tomorrow? Are revenue, risk, or opportunities hiding across your systems? I can pull that from Gmail. Okay. This sounds good. This sounds really good. Okay. If you want reply with one number, end of day control tower brief, plan review, I'm going to say I want all 3. And so that's what I'll say. I mean, I think this thing's pretty useful. I think if you want a managed version of this, you're probably going to need something like a single brain unified intelligence where you have everything connecting in and it has good memory and it's compounding with your organization. Plus you're constantly adding skills to it and you're adding other scaffolding to it. Again, the institutional knowledge is very important. So that being said, I'll catch you in the next video.