The Strategic Signal Squad

04:31:07.260 –> 04:31:36.780
Tori Jeffcoat: Yeah. So this QR code will take you to our staircase web page. If staircase is something that you wanted to learn more about and how we’re thinking about AI, we have a couple of really great pages around our Atlas agents. Our general customer. OS, if there’s more information that would be helpful definitely recommend visiting gainsight.com to learn more there. We also have some really cool ebooks. One that our CEO Nick Meta put together around. He thinks how he thinks about using AI using tools like staircase from that executive lens. So really great great. Read for anyone who’s a Cs leader and just wants to understand what that looks like.

04:31:38.070 –> 04:31:43.949
Julia Nimchinski: Amazing session. Thanks again, and we are transitioning to Zapier.

04:31:45.165 –> 04:31:56.739
Julia Nimchinski: Angela Ferrante heads up, enterprise marketing, and she’ll be taking us through strategic signal squad super! Excited, Angela, welcome to the show! How are you doing.

04:31:56.740 –> 04:31:59.580
Angela Ferrante: Hi, thank you so much. I’m good. How are you doing.

04:32:00.600 –> 04:32:06.359
Julia Nimchinski: Excited for this getting a lot of feedback from the community. So yeah, let’s dive into it.

04:32:07.320 –> 04:32:18.800
Angela Ferrante: Let’s jump on in so I am going to walk you through some of the no fluff. Real examples of agents that we use at Zapier.

04:32:18.810 –> 04:32:43.419
Angela Ferrante: I’ll start by going through a little bit of like framework on how we think about agents. The spectrum of automation, deterministic automation through agents, a couple of different frameworks for thinking about things, and then just dive right in. And I’m going to do something really risky which is to like, live, build and demo, the thing being the agents and see how that goes. So you’re gonna have to bear with me if technology gives us

04:32:43.420 –> 04:32:52.369
Angela Ferrante: problems, but no better way to show what we’re doing with our own agents than to actually show you them. So I’m gonna do that do that today.

04:32:53.500 –> 04:32:56.940
Angela Ferrante: So let me let me jump on in. I can share my screen.

04:32:58.320 –> 04:33:16.340
Angela Ferrante: So the agent, spoiler, alert, the agents that we’re gonna walk through are actually a team of 3 agents. And they are. We’re calling them the strategic Signal Squad, basically like a Pmm team that helps us do things around positioning and competitive messaging.

04:33:16.340 –> 04:33:40.020
Angela Ferrante: So I’ll get into that in a little bit. 1st of all, just Hello! I am Angela. I lead enterprise marketing at Zapier. I was previously a founder at a company called Laudable. We went through an acquisition, and I am now Zapier. I was a super fan and user of the product. Prior before that I was all in on marketing originally from the Boston area.

04:33:40.020 –> 04:33:53.540
Angela Ferrante: Hopefully, don’t have a leftover accent, but you’ll have to correct me if you are hearing one and a fun. Fact, I met a grizzly bear about 20 feet away from me in face to face in Wyoming a few weeks ago.

04:33:53.580 –> 04:33:56.080
Angela Ferrante: I don’t know if it’s fun or scary fact. But

04:33:56.410 –> 04:34:15.990
Angela Ferrante: there we go. So we are all hearing a ton about agents right? There’s a ton of noise complete, with even twitter trolls going around and commenting on things that aren’t true agents, right? This isn’t a real agent there is a lot that we’re all hearing in a lot of different platforms and and tools.

04:34:15.990 –> 04:34:43.560
Angela Ferrante: So I am here to help dehype this a couple of things. One is that what I’ll walk through today is something that can hopefully show you a way to build an agent today that actually works for you. It’s not like, Oh, cool! This is one pie in the sky that might be interesting one day, something that that works practically today. You also do don’t need to be an engineer just clarifying this like you don’t even need to have any advanced. No code skills.

04:34:44.131 –> 04:34:51.190
Angela Ferrante: I’m going to use Zapier agents, our agents product to walk you through this. But there are, of course, a range of tools you can use

04:34:51.400 –> 04:35:12.340
Angela Ferrante: really quickly. Zapier agents leverage the power of Zapier’s 8,000 plus integrations of app integrations to let you take actions. Automate steps do work on the web. Query your data sources, a whole bunch of cool stuff. We’ll talk through over 55,000 actions. You can take with over 8,000

04:35:12.340 –> 04:35:29.930
Angela Ferrante: work apps. So really broad coverage in terms of putting these agents to work on in the tools that you want, we have a bunch of of templates. There’s templates from Clairevo. There’s templates from Cat Gpt. If you follow either of those folks so you can head on over there. If those are interesting

04:35:30.830 –> 04:36:00.259
Angela Ferrante: and really quick before we dive in, I want to talk about this idea of the AI automation spectrum. This is a framework that has been helping us convey to customers who are eager to get into agent land like, what is the difference between a workflow and an agent? What are we really talking about? So we show in this slide? We’re showing on the left. You have a more traditional workflow, right? It’s entirely predetermined steps that follow a very predictable path.

04:36:00.509 –> 04:36:21.359
Angela Ferrante: If you use Zapier. One thing you may have noticed is that you can now add AI steps inside of a workflow, so you can manage a predetermined or pre-prescribed path, but start adding in different AI steps. And then we get into this agentic workflow, where there are more AI steps that are compounding and changed together.

04:36:21.360 –> 04:36:36.340
Angela Ferrante: Query, knowledge, writing messages, etc. And then on the right, we have, like a true agent, right where we’re combining. It’s fully AI. We’re combining natural language instructions, knowledge tools to autonomously complete

04:36:37.190 –> 04:36:37.660
Angela Ferrante: task

04:36:37.869 –> 04:36:52.570
Angela Ferrante: that are more complex and that that the agent is using its discretion and reasoning power to complete and get to a pre prescribed goal. So I think the thing that I found really interesting is that there’s a different role, or there’s there’s different

04:36:52.570 –> 04:37:09.160
Angela Ferrante: type of tool for for each job. Sometimes you may want something to follow a preset path and be really prescriptive, and sometimes you just want it to figure out how the heck to get to an outcome. Most organizations that we see need solutions across this entire spectrum

04:37:11.474 –> 04:37:36.020
Angela Ferrante: one other fun. Fun diagram here is so at Zapier we are now. We published about a month ago that we’re at 89% daily AI adoption across our team. I think we’re now something like 96, or inching towards a hundred. We actually have more active agents than we do employees at this point and one way that we like to think about. This is with the in an org chart. So this is a

04:37:36.020 –> 04:37:45.109
Angela Ferrante: a small sampling from our our go to market org, where we think about different phases of of go to market, engage, sell, deploy, manage, etc.

04:37:45.110 –> 04:37:54.480
Angela Ferrante: And these are some of the the use cases that we think about. Within. Within these categories we are starting to publish more of these org charts and and templates

04:37:54.480 –> 04:38:18.050
Angela Ferrante: supporting them. So we launched Zapier for Hr. Where we have a ton of cool AI use cases within within the Hr function, recruiting, onboarding, etc. And we’re we’re starting to build up to marketing as well. I’m doing an ama session. I’ll talk about that in a little bit for anyone who wants more on how we think about this and AI fluency across across the org.

04:38:18.110 –> 04:38:43.629
Angela Ferrante: One last thing I want to call out on this is that we believe that the people closest to the problems are best able to solve them, and agents are a really great way to enable go to market folks on each of your teams to actually solve, build solutions for their their problems versus having to use out of the box solutions or solutions created by folks who aren’t actually in touch with their tech stack their work, streams, their their processes, and so forth.

04:38:44.960 –> 04:38:58.070
Angela Ferrante: Okay, so let’s get into it. We are going to focus, or I’m going to walk through this strategic signal squad this team of AI agents that helps with Pmm tasks

04:38:58.480 –> 04:39:28.059
Angela Ferrante: and setting the stage here. So at Zapier we have about 12 product marketers across a couple different functions, and we are moving like many of you, moving up market. Which means that our Pmm bandwidth and research bandwidth in particular, is pretty limited. At the same time, the landscape, as we all know of AI tools, what possibilities there are is moving really, really fast.

04:39:28.060 –> 04:39:39.220
Angela Ferrante: So we’re also launching along with that, we’re launching products really quickly. And so our bandwidth is is pretty limited. These are some of the agents that help us move faster. You know. Classic do more with less.

04:39:39.220 –> 04:40:09.000
Angela Ferrante: And you can’t see these screenshots. But don’t worry. I’ll go into an actual screen share on these shortly. So the agents we’re gonna walk through. There’s 3 of them with a bonus. So the 1st one is a competitive or I’m sorry a company insight extractor, and the idea is that you may want to run this on your own company to just get a rundown of what’s the public facing view of of how we’re coming across our positioning our Icp, our value prop our differentiators.

04:40:09.270 –> 04:40:21.443
Angela Ferrante: Or you might wanna run this on a competitor. There’s a bunch of different ways. You might use this also a prospect or a customer. If you’re looking for a quick brief before a call to understand their business better.

04:40:21.870 –> 04:40:40.280
Angela Ferrante: The second one is all about competitive. So we have a single shot sort of robust competitor messaging differentiator. That’s a mouthful that you can run one time to run a detailed report. I’ll show you what that can look like of competitive messaging versus your own.

04:40:40.280 –> 04:40:55.949
Angela Ferrante: And then I really like this one. This is the one I’ll do more of a deep dive on is the weekly competitive Update investigator. The idea is that this will go parse all of the competitive intel that is coming out, or all the Intel about your competitors.

04:40:55.950 –> 04:41:15.079
Angela Ferrante: Press releases news product launches every week and give you sort of the. So what what should we do about this? Right. Maybe they identify some white space in in the market, or they identify an area that is encroaching upon your Icp. And positioning that the competitor launched

04:41:15.820 –> 04:41:41.920
Angela Ferrante: the last agent in this squad is our product launch messaging and positioning writer. Again I mentioned, we’re doing a lot a lot more launches and starting to enable AI to give us drafts of more of the the Pmm launch work. So things like messaging value prop positioning new product for new products or features, you could add an email, copy, copy, social media launch, copy and so forth.

04:41:42.724 –> 04:41:48.280
Angela Ferrante: Let me pause any any questions, Julia, anything you want me to dive deeper on before we we jump on in.

04:41:48.640 –> 04:41:51.850
Julia Nimchinski: Let’s jump on in. And yeah, we’ll address it in the end.

04:41:52.170 –> 04:42:03.320
Angela Ferrante: Great. Let’s do it. So the I’m gonna show you one by one each of these and and sample outputs, and then I will do a deep dive on this agent. 2 bonus.

04:42:03.370 –> 04:42:04.100
Angela Ferrante: So

04:42:04.840 –> 04:42:32.350
Angela Ferrante: company insights. It’s extractor. This is here is that agent. I will share all of the template links. They are also in the slide. So if you want to create your own version, you can actually just go click, use this template, do a few quick configures and and use it. So this is the agents product inside of Zapier. We’re basically giving it some instructions about who it is similar to a prompt, but it has more access. It has access to different tools.

04:42:32.350 –> 04:42:37.960
Angela Ferrante: Then it can take more more actions, all in one versus a single back and forth prompt.

04:42:38.305 –> 04:43:02.629
Angela Ferrante: So you’re basically asking it to ask you as the user questions around the homepage that you can enter in different information, it may go back and forth with you and ask you for more for more. Intel and then here is an an example. So we ran this on on a competitor. And it actually will generate a Google Doc. You could have it share with different teams. You could have a ping in slack.

04:43:02.690 –> 04:43:14.519
Angela Ferrante: But the the Google Doc output shares a range of product and positioning key features, customer and community notes the various value props and

04:43:14.970 –> 04:43:16.560
Angela Ferrante: differentiators.

04:43:17.350 –> 04:43:24.199
Angela Ferrante: So that’s this 1st agent, the second agent. Let’s get into the competitive messaging differentiator.

04:43:24.819 –> 04:43:44.270
Angela Ferrante: So this one is again, you’re gonna we’re gonna give this this agent a role and tell it what its goal is, and what the what the output is. As you can see, there’s a lot more complexity in here than a traditional prompt. And it has is taking multiple steps on its own with access to multiple tools.

04:43:44.270 –> 04:44:00.150
Angela Ferrante: So I won’t go through every single one of these. But there’s a whole bunch of steps where it’s asking. It’s gonna ask you, as the user to copy paste different things from your competitor, share more about your own problem and solution. It’ll do a tear down.

04:44:00.150 –> 04:44:08.399
Angela Ferrante: it’ll identify white space. And the you can run this on demand at any any time for any any competitor.

04:44:08.400 –> 04:44:33.359
Angela Ferrante: And here’s what the output looks like. So again, we ran this against some of our common competitors out there, and it will even create we have in this one. It’ll even create these matrices or sort of comparison charts to show you functionality for different, for the various different products, and how they compare. And then it’s gonna show you any areas and gaps in app positioning, messaging, or or product offerings

04:44:33.390 –> 04:44:49.769
Angela Ferrante: so pretty robust. And all of this you can you can customize. So perhaps you want to add a step? 9 that you want to have it. Look at. The future state of of product features, and assess what that might look like you can add this, add this in here.

04:44:50.700 –> 04:45:15.029
Angela Ferrante: A variant on this competitive mess or competitor messaging differentiator is maybe you want a weekly update. Right? Let’s say you want to keep your eye on these competitors and get any relevant news launches updates in the market that might make you change your your competitive battle cards, your positioning or your messaging more generally.

04:45:15.664 –> 04:45:23.780
Angela Ferrante: This is a template that for an agent that will do that, and we’ll actually send the slack message to your team every single week.

04:45:23.910 –> 04:45:52.620
Angela Ferrante: so you can schedule it when you want. This one also uses perplexity, and if many of you are using perplexity, but we’re starting to see it gain. Gain steam against Google. Chat Gpt, great for research tasks. So this is a template where you can drop in your company, and every week it’s going to start with a perplexity, completion, that is, or search that is going to look for updates across all of your various competitors.

04:45:52.730 –> 04:46:21.559
Angela Ferrante: We then put this analysis framework, that which is the way that we want want Asapier, that we want this agent to think about the various components of competitive comparison. So we looked at direct feature comparisons potential for market disruption. Where is it infringing on our category? Right? What advantages might they be developing. And then any market gaps or competitor or customer pain points.

04:46:21.850 –> 04:46:33.940
Angela Ferrante: We’re also saying, Hey, are there actual partnership or integration opportunities? Sometimes you have a. You know, the co-opetition type type folks out there that are sort of competitors sort of sort of partners.

04:46:34.080 –> 04:46:43.630
Angela Ferrante: And then strategic implications. Right? We all want to know what’s the. So what? What should we actually change about our own messaging? What do we do with these new, these new updates?

04:46:43.710 –> 04:47:07.449
Angela Ferrante: And then we have it. Send a slack message every single week that will will lay some of these things out and give us some suggested action items. So a lot of detail in here, a couple of things I want to show you. First, st I want to show you what it looks like to actually use this template and build your own agent, and then I’ll show you what the outputs for this one this one can look like.

04:47:07.890 –> 04:47:09.990
Angela Ferrante: So if you click.

04:47:10.150 –> 04:47:10.840
Angela Ferrante: But

04:47:11.050 –> 04:47:22.100
Angela Ferrante: if you click, just use this template. In this case it uses perplexity and slack. So you just would connect your own perplexity. Api key plus your slack account in here.

04:47:22.210 –> 04:47:24.460
Angela Ferrante: and you’d click, create.

04:47:24.570 –> 04:47:38.360
Angela Ferrante: and voila! Here is your agent. So any of this you can customize. You would complete this with, you know. Enter your company. Let’s pretend we’re open. AI, you would fill this out. You can also adjust

04:47:38.360 –> 04:48:02.950
Angela Ferrante: the analysis framework for your own market if you care about different things if you want it to do different things, and then you can select your your slack channel. And this is where you’re giving the agent different tools. So all of our, you know, 7, 8,000 different apps. You can add the capability for it to do different things. Maybe you want to add it, have it, create a Google sheet for you, or add a row to a particular

04:48:02.950 –> 04:48:26.090
Angela Ferrante: Google sheet. Maybe you want an Asana Clickup or monday.com task added, and assigning it to a specific person, could even route to a particular person depending on the nature of of the task and the and the topic. A whole bunch of things again, of course, will not go through all thousands of them, but many different tools and capabilities. You can give your give your agent here.

04:48:26.690 –> 04:48:46.570
Angela Ferrante: From there you get to go start testing it. So just so we don’t have to wait. I populated this one that fully assumes we’re running this for Openai, comparing against anthropic Google, Meta and Deepseq. Obviously, lots of launches, lots of things happening in that space every single week.

04:48:46.570 –> 04:48:56.839
Angela Ferrante: And here is the I ran a quick test on it. It 1st it pulled from perplexity, and then it ran all of this other other analysis, and and would send it in a slack message.

04:48:56.840 –> 04:49:02.769
Angela Ferrante: And here’s what this output looks like. So it’s telling us it’s giving a quick summary telling us

04:49:02.990 –> 04:49:13.800
Angela Ferrante: anthropic won a legal battle regarding AI training on copyrighted materials. Some of you may have seen that talking about the positioning of Deepseek’s latest updates.

04:49:13.940 –> 04:49:23.270
Angela Ferrante: lack of updates from Gemini and then Meta, winning a copyright lawsuit. So that’s like the quick executive view as a leader that you may want to understand.

04:49:23.270 –> 04:49:44.909
Angela Ferrante: And then we break this down or it breaks it down by each each competitor. So it’s talking about what happened with Gemini. It’s also going to give a threat level of the updates per per week, right to sort of tell you how severe, what what you should keep your eye on, as well as the particular opportunity in place, and of course it will cite its sources.

04:49:45.090 –> 04:50:13.689
Angela Ferrante: so it’ll go through all of these and then give you these strategic takeaways. So we’ve got immediate action required. Let’s take a look at this deep seek and anthropic updates. Both could impact our competitive positioning and data strategy. It’s also giving us recommended actions. Maybe we need to do some more benchmarking against deep seek an example I’ve seen before. Come out of this is, update the battle card. You could even add the battle card

04:50:13.690 –> 04:50:38.029
Angela Ferrante: as a if you have that in like clue or crayon, or even in just a Google Doc, you could add that as a tool inside of of your agent, and it could directly interact with with those battle cards or pull from the battle card data to suggest, hey, you should add a bullet point on this particular team, so you could be proactive before sales comes running, knocking for asking for for additional battle battle cards

04:50:38.090 –> 04:50:41.610
Angela Ferrante: based on the latest news. You’re already ahead of ahead of the curve.

04:50:42.611 –> 04:50:48.619
Angela Ferrante: And then it just shows a little bit about what it’s gonna do in the the following week, and it gives you this

04:50:48.860 –> 04:50:57.379
Angela Ferrante: of the disclaimer but a little bit of a caveat, that this is a an agent calculated report or an agent generated report.

04:50:59.010 –> 04:51:10.369
Angela Ferrante: So this is this is the agent. You can configure this to do any time you want any frequency you want. I like weekly or monthly for these these competitive agents.

04:51:11.952 –> 04:51:20.530
Angela Ferrante: So that is that one I know. I went through these really quick again. You can steal all of the templates on this

04:51:20.530 –> 04:51:45.220
Angela Ferrante: slide here the 3rd one. I’ll just zoom through this last one. We’ve got the product launch messaging and positioning writer. So we are aiming at Zapier’s Pmm teams to get 1st drafts. I like to call them shitty 1st drafts, not because they’re shitty, but they give us the power and sort of like just ability to let go of the stress around the blank

04:51:45.220 –> 04:51:56.600
Angela Ferrante: page problem and get something drafted really fast. We use this one on our new product and feature launches from, particularly from tier 2, 3, and 4 launches on Pmm.

04:51:56.620 –> 04:52:00.129
Angela Ferrante: And here is how you pull this one up.

04:52:01.440 –> 04:52:07.660
Angela Ferrante: Let’s see, cool so it will. This is one that we can run on demand

04:52:07.660 –> 04:52:31.870
Angela Ferrante: and that you can interact with. So we have a whole bunch of steps here that it’ll tell us that that it will ask us for information when we interact with it, and we can also feed it. More information on our brand guidelines, on our messaging house, our standards, our positioning, our Icp, you can really give it or direct it to all of the information that maybe updates regularly. That’s really critical and

04:52:31.870 –> 04:52:36.069
Angela Ferrante: foundational to developing, messaging, messaging, content and

04:52:36.440 –> 04:52:51.499
Angela Ferrante: and updates. So the final app output on this one is that you’re gonna get a point of view and messaging framework ready for executive review. That will be for a particular product or or feature.

04:52:52.310 –> 04:53:15.129
Angela Ferrante: Here is we ran this just prior. We ran this for our agents product so a little Meta. But this is the product that I’ve been going through. We’re working on an up a new launch for for agents later this this summer with a bunch of new features, and we had it. Write a draft point of view draft on our messaging. You can see it lays it out in terms of the challenge, the consequence, the future state

04:53:15.350 –> 04:53:39.529
Angela Ferrante: and target benefits. You might have a different framework that you want this to follow. Like, say, you have your tier. 2 launches follow a very particular structured framework. And you even want to draft an email copy for customer launch announcement. You could incorporate all of that into into this step and or add a step to to make it really custom.

04:53:40.973 –> 04:53:41.899
Angela Ferrante: The last

04:53:42.489 –> 04:54:05.500
Angela Ferrante: so that that’s it for the agents again. I know I went through these super quickly. They’re all very customizable. They’re all stealable. You can get one of these working, and I will. I will issue this challenge to all of you in under 10 min and have something live that that’s that’s working for you. So really simple to set up with the power to add a whole lot of customization.

04:54:07.650 –> 04:54:31.070
Angela Ferrante: I will open up to questions if we have any. Before I do that. I mentioned earlier this AI fluency model that we’re thinking about at Zapier. So our Hr. Team published the framework for how we think about AI fluency across the org and upskilling folks to be AI, native AI first, st and how they think and and create solutions for their their work day to day work.

04:54:31.190 –> 04:54:41.010
Angela Ferrante: We have one on marketing. This is just a teaser of of the framework that we think about things from unacceptable, capable, adoptive to transformative.

04:54:41.499 –> 04:54:51.930
Angela Ferrante: These are some of the the high level. But we have a lot more on impact behaviors across specific marketing functions demand Gen. Growth. Pmm, to make it really concrete for folks on our on our

04:54:52.100 –> 04:55:05.410
Angela Ferrante: almost almost 70 person marketing team. What is, you know, the the behaviors and the skills that they should be? They should be aiming for when it comes to AI we are going to host an ama on this.

04:55:05.530 –> 04:55:27.290
Angela Ferrante: We’ll have a special guest not yet announced, but we’re doing an ama on our AI fluency, giving away our template for the AI fluency model so that we can help you if you’d like workshop your own. And that is on Thursday, July 17.th This QR. Code should take you to a Luma event. If you want to attend, we’d love to have any of you.

04:55:28.440 –> 04:55:30.369
Angela Ferrante: that’s all. I got questions.

04:55:30.370 –> 04:55:33.889
Julia Nimchinski: Super insightful. Thank you so much, Angela. And

04:55:34.870 –> 04:55:39.239
Julia Nimchinski: to address like the question we actually address in every session today.

04:55:39.770 –> 04:55:52.989
Julia Nimchinski: and we chatted about it a little bit behind the scenes. But what your advice for all executives watching Gtm leaders in terms of transitioning to a gentic scaling

04:55:53.210 –> 04:55:55.609
Julia Nimchinski: this new operating system.

04:55:55.750 –> 04:56:02.190
Julia Nimchinski: And given the fact that you have you mentioned 96 over 96% AI adoption, correct.

04:56:03.340 –> 04:56:11.770
Angela Ferrante: yeah, yeah, so we we actually open source our own playbook for AI adoption at Zapier there’s a a link I can always link to folks

04:56:11.830 –> 04:56:17.470
Angela Ferrante: link folks to that. There’s a link in our on our website. It talks about the how we hit that 89%.

04:56:17.470 –> 04:56:42.349
Angela Ferrante: And we broke it down into a couple of different frameworks includes the spectrum of like, what are the categories to be thinking about covering tech, stack governance, use cases and culture. Those are the really the big components that we we think about. And within that we open, sourced, even like our our original slide deck. When our CEO Wade called a code red around AI year and a half to almost 2 years ago now.

04:56:42.370 –> 04:56:59.180
Angela Ferrante: and made this an org wide priority. What I’ve seen work really well is one just naming that urgency and and priority so that everyone across the org knows that it’s it’s a top impacting, impactful priority for everyone to be working on.

04:56:59.180 –> 04:57:26.490
Angela Ferrante: and then providing the space and time. I think Wade says if you do just one thing, it’s run hackathons and giving people that that space outside of their day to day work to be able to to experiment with with different products. There’s a lot of tools out there and that can get is exciting, but can get overwhelming. So you just need some space to 1st be bad at things to get over that hump, and then the other thing I’ll go back to that, I said. This earlier is.

04:57:26.770 –> 04:57:53.689
Angela Ferrante: I really think everyone is a subject matter, expert in their own work function and their own workload, and as such they are best suited to solve solve the problems right? Like they’re closest to it. They know how to do it. But historically, that’s been really hard to do if you’re not also an engineer. So that that’s where I think, playing with workflows with with AI steps and agents. Full agentic

04:57:53.760 –> 04:58:02.599
Angela Ferrante: steps and and tools is really good, like they just need the time to to experiment and and be close to the work and and dive on in.

04:58:04.190 –> 04:58:10.839
Julia Nimchinski: Next question here. What’s 1 place you tried to automate signal tracking, and it didn’t work.

04:58:11.160 –> 04:58:12.700
Julia Nimchinski: What did you learn from that?

04:58:12.930 –> 04:58:19.360
Angela Ferrante: Hmm! I think. You know, you’re only gonna get and interpret so much from

04:58:19.400 –> 04:58:48.269
Angela Ferrante: what competitors are publishing, I mean, surprisingly depending on your particular space and competitors. You can get a lot of information. I think an area that is a sort of hidden gem for competitive intel is job descriptions right like a Jd. That a company is putting out oftentimes has a lot about their strategy. And there, so we didn’t have that in in this particular agent, but I think, adding that in would be a wise step in terms of something that didn’t work.

04:58:50.520 –> 04:59:17.890
Angela Ferrante: you know. I don’t think you can have this yet. Make all of the updates in your battle cards for you. Right? There’s like, if you think about the steps it’s gathering the Intel AI can definitely do that. It’s making recommendations and drafting things. AI can definitely do that. And then it’s actually going through and having the discretion. And to make the updates. I don’t think that AI without a human in the loop at this stage is is equipped to do that. So I think

04:59:17.890 –> 04:59:26.969
Angela Ferrante: some of the trial and error we’ve seen there, where we’ve like pushed the boundaries and figured out what is and what isn’t viable for for AI to accomplish.

04:59:28.130 –> 04:59:31.880
Julia Nimchinski: One more question here. We track competitors manually. Now.

04:59:32.050 –> 04:59:35.889
Julia Nimchinski: how did you decide what signals were worth automating? First.st

04:59:36.916 –> 04:59:39.629
Angela Ferrante: Will you see the 1st part again we track. What was that.

04:59:39.670 –> 04:59:41.710
Julia Nimchinski: Competitors, manually.

04:59:42.295 –> 04:59:45.050
Angela Ferrante: Yeah. And then what? The second part of the question.

04:59:45.050 –> 04:59:51.160
Julia Nimchinski: Yeah. And the second is, how did you decide what signals were worth automating? In the 1st place.

04:59:51.810 –> 05:00:16.470
Angela Ferrante: Oh, I mean, I think it starts with the fact that you’re doing it manually is the perfect start, right? It’s start with what you are already doing what the human would do. That’s a really great blueprint for looking at. What can you automate? So, for example, if a human is tracking competitors manually, and they’re going in once a week or once a month, depending. If you have an Fte or just partially dedicated team, if they’re looking at

05:00:17.320 –> 05:00:32.409
Angela Ferrante: press releases website changes and job descriptions, for example, great. That’s your your template for what you can give an agent or a workflow to to automate this work and have AI do it for you. So I would really start there and then.

05:00:32.460 –> 05:00:49.140
Angela Ferrante: Such a dumb tip. But ask AI right in just a standard chat, gpt or plot, or you know whatever you use, give it what you’re already doing. Give it your manual process, and say, how can I make this better given this, we’re really missing, you know, we really want to know more about their

05:00:49.150 –> 05:01:08.680
Angela Ferrante: employee sentiment or something, you know, whatever it may be. And maybe it suggests that you start tracking steps in Linkedin or or employee posts on Linkedin or the the team size that Linkedin shows on the company pages, so it may give you more ideas for things that you feel like you have gaps in in terms of your your competitive intel.

05:01:09.470 –> 05:01:11.909
Julia Nimchinski: Thank you so much, Angela. Again

05:01:12.170 –> 05:01:18.869
Julia Nimchinski: everyone follow Angela on Linkedin. I’m assuming, and do attend Zapier’s next event.

05:01:19.370 –> 05:01:29.700
Angela Ferrante: Yeah, yeah, please, please join me or connect with me on Linkedin. And then, if you’d like to join our our ama on AI fluency. You can either send me a message or go to the QR. Code.