The Evolving Role of the Data Analyst

Ben Jones: All right, hello everyone. Welcome back to the Powered By Data show. I'm your host, Ben Jones. And this is a really special episode actually. It's a paid sponsorship from Hex. We're super excited about engaging in a broader partnership with Hex. We've never done this before as a company. It's the first time in our eight year existence that we've ⁓ taken on a tool partnership like this, but we are just really big believers in what Hex is building and what they have as just an amazing AI first analytics tool.

So joining me on the show today is Rachel

hi, Rachel. Let me tell you a bit about Rachel, then she'll give her background herself. So she's a practitioner. She's been at EY, Slalom, Amplitude. Now she's a product evangelist at Hex. I interacted with Rachel ⁓ a few weeks ago. I read an article she wrote about what the data analyst, what the future is for the data analyst, and it really caught my attention.

Rachel Herrera: Hello.

Ben Jones: So I reached out to her and said, thank you for the article. It really resonated with some conversations I was having with my students in the classroom at the University of Washington Foster School of Business where I teach. so Rachel and I had a little conversation in DMs on LinkedIn about it. And now I get to have her on the show. So I'm super excited about that. So welcome to the show, Rachel. Why don't you tell our listeners kind of a bit about yourself.

Rachel Herrera: Cool. Yeah. Thanks Ben. Happy to be here. So about me, I've been a data nerd for, gosh, 12 years, 13 years. don't know. My original degree is in marketing. So don't know how ended up in data, but I think that's sort of how it goes. If you love it, it just like grabs onto you you can't let it go. And I've really loved sitting in the middle between the business and the technology. And like, that's where I think I wanted to be a teacher at one point. And I feel like that's really...

what I lean into when I think about data is it's, yes, we need to create these trustworthy insights. We need to build beautiful dashboards. What I've spent a lot of my career doing is teaching others how to think and do and ask good questions with data. Because I think that's the big gap that this profession can help fill. And it's one that I'm most excited about filling, especially with AI and sort of the future of this role, I think, lets you do a lot of that stuff a lot more, whereas in the past you were sort of stuck.

writing lots of SQL and cleaning data and doing the things that maybe weren't as a business savvy business facing, but always important. How can we sort of maybe push that to the side and get to do some of the more like storytelling aspects that I've really enjoyed in my career. Because in my time at Amplitude, especially I was in professional services. like literally sitting in between the technology and the business and saying, Hey, we want to do these things with this tool and this data. How do we do that?

and learning a ton about, maybe this is a lukewarm take hot take. don't know. everybody more or less has kind of the same problems. I think if you're a business and you're struggling with our data is messy or we don't know what to do with it, or we're trying to expand access or our analysts are bottlenecked - believe me, you are not alone in that frustration and, looking at a problem like that in so many different ways, you sort of realize that the only way out is through.

Ben Jones: Okay.

Rachel Herrera: and getting to the action, doing the work is the best way to sort of figure out how to bring those solutions to bear that you're trying to bring to the business. So that's what I'm most excited about talking about today, because I think that's a place that analysts in the future, especially with AI, are just going to be the change agents that are necessary to make that real.

Ben Jones: Okay, so tell me more about that. So, you know, there's a lot of anxiety out there, people who are studying or practicing the craft of data analysis.

They're really worried that somehow now AI can do so many things that they used to do. Like you mentioned, creating SQL queries, building dashboards, sharing reports. So if AI is making some of those tasks easier and easier, they're sitting there thinking to themselves, well, what am I going to do then? What is my role? What is my value add? So talk to me a little bit about that. You wrote about it in your article.

Rachel Herrera: Yeah.

Ben Jones: But what do you see as the key contributing factor or the role really for a data analyst in an organization today? How is it changing? Is it better? Is it worse? Kind of talk me through all of that. Like what does a day in a life look like now that these tools are becoming more widespread?

Rachel Herrera: Yeah. And I think it's better in a lot of ways. I think it's also more challenging in a lot of ways. I would say the biggest change from what I remember in my days of, I get a question. comes in the queue. I find the data associated with that question. I write a SQL query to answer that question. And then I build some sort of reporting or story to say, Hey, here's what I learned. And here's what I think you should do. Or here's the facts and figures and you go make whatever decision you'd like. And.

We all know that that turn wasn't a one-time thing. was, I gave them something. They said, that's not quite right. Actually, I thought of something else. Could you add this into it? wait, I forgot to tell you about xyz thing. And so was like a very iterative process. And that I still think is like a very real thing that analysts are going to have to continue to be good at. Because somebody mentioned this at Hex the other day, and it's like my favorite thing now, is that data is a soft skill.

Ben Jones: Yeah.

Rachel Herrera: Yes, there's the technical pieces to it, but ultimately the things that I think make it useful to a business is a human being able to relate and understand and have that sort of dialogue with another human and figure out what are they trying to get at? are they trying to understand? Now, what I think AI is going to let an analyst do is be able to teach an agent some of those things that they just know inherently about what the business cares about, how the data is structured.

The things that maybe are less interesting in a conversation about what am I trying to do with data and more of something that you would do sort of when your head's down looking into a problem. Teaching an agent how to do those things then lets that agent multiply your value to the business because now they don't have to come to you. They can come to the agent that you're training and they can ask those questions. They can do those multiple turns, figure out, okay, this is what I'm actually trying to get at. Okay, great.

Ben Jones: yeah

Rachel Herrera: And what happens is now that individual has gotten their question answered with data. And the analyst that trained that agent can review those answers and make sure that it's doing what it's supposed to be doing. So that kind of leads me to my first point is like, think analysts will become curators to allow that loop to be a) trusted, but also b) really fast. And so that businesses can get a sense of like, okay, this data is useful. It's helping me make these decisions. I feel good about the answers it's giving. Or if I don't,

Ben Jones: Okay. yeah

Rachel Herrera: I have a recourse and a way to flag that back to my analyst whose responsibility is still deliver trusted insights to the business. But now this is like a new delivery mechanism as opposed to let's cut them out completely because we don't quote unquote need them anymore. In order to make that loop valuable, you absolutely need them more than probably anything else outside of the data itself, of course.

Ben Jones: Well, because that question of trust, you mentioned that word, that's the important thing, right? A lot of organizations are sitting there saying, can we trust any of this? So how do we know if it's accurate? How do we know if it's real? Did it get just?

Hallucinated is it something made up? So this is a question of course that organizations have been asking for good reason I mean, know over the last few years LLMs have been evolving and improving but in the early days especially they were pretty bad and so as Remarkable as they were they would continually make pretty blatant and egregious mistakes Of course, they're improving but we still always have that question. Wait a second. Is this accurate? Is this based in real data? There's the additional question of well, can we even trust the data itself?

independent of the LLM. So talk to me a bit about Hex and what you all are doing to try to give some tools for analysts in the business to understand how trustworthy the insights are and also then to be able to make improvements to the overall system and architecture to just increase trust over time.

Rachel Herrera: Yeah, because when curation becomes the responsibility of analysts, you know, they are responsible for telling the agent what's important, what data means, what the business cares about. ⁓ that needs to be a workflow that is built into whatever product they're using to push the insights out. And so I love that Hex is thinking about this from a workflow perspective and not just a feature functionality. Like it's, it's not, Hey, here's a thing. You could do whatever you want with it. They're being.

pretty prescriptive and intentional about how it's best to let an analyst work through and manage and curate context through this thing we call Context Studio. And in Context Studio, you can manage context across a bunch of different layers. And this is the other interesting thing that I think Hex does that I actually really align with is there's a bit of a bifurcation happening in the data world when it comes to how should agents talk to data. I'm sure you've heard it.

to semantic model or not? Do we do all in 100 % on building out these highly governed semantic models that give the agent these really tight guardrails and you cannot go off road with it? It can only answer the questions that have been sort of pre-modeled. I don't know about you, but that sounds like a past that we've already lived through. Where we've got dashboards that have filters and that's all you can do with them.

Ben Jones: Mm-hmm.

Rachel Herrera: And any sort of deviation from that requires like intervention from the data team. And there's pros and cons to that. Absolutely. What Hex, I think, sees the future as is there is a world where you need these governed answers. So in context studio, you can create semantic models just like you would any other tool. You can sync them from external tools, which is also great. One thing I love about Hex is it's very like interoperable. It's not precious about anything. It's like, hey, we do this workflow the best.

But if your data lives over here and your semantic models live over here and you used to have stuff in GitHub and you want to connect all those things, that's where I think Hex really shines. So you can do the semantic model route, or you can also curate the context for the more exploratory stuff. So workspace guides, workspace context, all of the little bits and bobs that maybe don't live nicely in a data warehouse, but also the metadata and the column.

descriptions and all of the kind of warehouse maintenance that needs to happen. And the agent sort of reads from all of that, including even like more stuff, like the actual work that you've created in Hex becomes context. And so it like compounds, the more you use it, it just gets smarter and it's just, it's really exciting to see because I think that's the future. That's, that's when you think about self-service, it's not just the what happened questions, which I think semantic models serve.

really well. That's their lane. But then you ask the why. Why did revenue go down? Why did the marketing campaign not be successful as it was last year? And those are very difficult. And those are things that can't operate in a semantic model most of the time.

Ben Jones: Yeah, there's that rigidity of the semantic layer and the need to fully specify everything, which I think organizations know is simply not realistic. And then of course, there's the opposite extreme of the wild, wild west where you don't even define anything. And then, you know, everyone's all over the place. So I like where you're coming with this, you know, there's got to be sort of a balanced perspective in the middle where you have some very well-defined definitions and descriptions and metadata and metrics all called out.

Rachel Herrera: Yeah.

Ben Jones: and you have also the ability to kind of evolve that over time. So you mentioned this feature context studio and you know, I don't know about you, but I love to kind of see things. This will be a little tough for the people just listening to the podcast. But if you're watching along on YouTube or on Spotify video, then that's great. So you can check this out because what Rachel is gonna do is kind of give us a quick little.

preview of what, well, it's live. mean, this was launched recently, right? I want to say maybe a few months ago, context studio. Okay, great. So this is out there. You're seeing people use it. Your clients are using it. You're using it at Hex as well, your data team. So here we go. So talk us through what you're seeing here, Rachel. And then for the folks listening, you can kind of imagine ⁓ how Hex is assisting a workflow here that is super critical.

Rachel Herrera: January of this year,

Yeah. And this is my little test environment. So it's just me tinkering around in here. So, ⁓ imagine sort of that you've got hundreds of agent conversations. And if I'm an analyst or a data leader or even a data engineer or an analytics engineer, anybody that's sort of responsible for our agents, giving good answers and how is our adoption trending of, ⁓ as rolling out agentic or AI analytics. ⁓ this sort of becomes like my home

And so.

I can come in here and get a bird's eye view of sort what's happening across my organization, and I can drill into individual, we call these threads or conversations depending on where they come from. ⁓ And then what's even cooler is, let's say I click into one of these threads, we also surface these things called warnings. So the agent responded to this question, I just simply asked it, you what is our fiscal calendar? And it didn't have any of that in its context.

There was nothing in it telling it what our fiscal calendar was when it started. And so rather than giving a bogus answer, ⁓ what Hex has done here is really harness this agent and say, you need to operate within these guidelines that have been created for you by the individuals who are administering Hex. And if you run into some issue or some limitation or some confusion, surface that. And then it also gives us a suggestion on how we could go and resolve this issue.

So this is a really nice workflow, especially if you are rolling this out across a wide swath of people and you need to maybe filter based on, we'll actually start generating topics for you as the conversations come in. And then any warnings that surface, ⁓ even filtering it by user or by the different roles. ⁓ So a ton of flexibility to be able to come in here and just sort of see what's happening.

Ben Jones: So talk me through that workflow. Let's pretend there's a scenario. ⁓ A business user interacts with an agent. It causes there to be this warning on the back end. So how does the data team know about it? How do they respond to that? And I guess most importantly, from your point of view, Rachel, how is that an empowering thing in terms of giving the data team a new superpower?

Rachel Herrera: Yeah, because when you think about these questions happening maybe elsewhere or outside of a tool that's sort of like re-ingesting them, ⁓ data people love data about data. And so ⁓ it becomes really difficult to understand, well, okay, how is all this performing? Because ultimately you're gonna be responsible for the quality of the responses. ⁓ Because ideally my objective or at least my goal or

belief and wish for data professionals is to be sort of the stewards of AI on data at their organizations. And one way that they can make sure they can do that is be able to say, how do we know it's working? And so this really, I think, enables that workflow. So let's say I'm not Rachel. I'm a different user. But I lead a data person or lead a data team. And I come in here and I say, OK, how are we trending? How is our conversation volume trending? And I realize that, hey, we've got a warning here.

on one of these conversations. If I click into that, like I mentioned before, I can come in here, see this warning and be able to triage sort of, okay, I can see this entire conversation. I can see the thread. I can actually see some of the thinking that the agent did. This is pretty lightweight, but I can jump into a more complex one. ⁓ And then I can understand, okay, it's telling me that I should add documentation to my workspace rules. ⁓ And I should be more specific about when the fiscal...

calendar start date is. And so just by that suggestion, I could come back here, go into the more free text like guides and workspace context, which is what it's telling me I should go update, and add some of that here. And this is all of the

unstructured context that the agent can use and that the data team is sort of responsible for also creating and managing.

Ben Jones: more like a system prompt for the agents to refer to and give them that background information.

Rachel Herrera: Yeah.

Exactly. And what this looks like today is one thing, but very soon, I think actually within the next few days, we're launching a feature that makes this whole experience a lot more agentic and a lot more maybe just a human in the loop approval, as opposed to, I have to click into this to go see, and then I have to copy paste something or change something somewhere. Because if we go back to this example here,

Ben Jones: Mm-hmm.

Rachel Herrera: ⁓ What we're releasing is like a, we're calling it a review agent, but also suggestions. And so what the review agent would let me do is rather than just simply surfacing this suggestion, I'd actually see a button here that says, go into the suggestion flow. It'll take me into suggestions and I can actually see all of the other conversations that encountered a similar challenge and a similar confusion of maybe not understanding when our fiscal calendar started.

And so again, as a data person, I can go and size the impact of this gap in our context and say, wow, 15 people have encountered this and two of them are in leadership. That's really important for me to go fill that gap now. And rather than me having to click into guides, paste a thing and do a thing, I'm actually going have an entire sort of feed that I can just click into. The agent will surface what exactly the new guide

terminology or text should say. And then I can just hit a button and publish it directly from there. And then I've cleared that backlog of the 15 pieces of confusion so that when those users come back and ask another question, now they get a more trustworthy answer. And so I it really speaks to like data is never done, but also context is never done because people have new questions, interesting questions, and it's going to be on the data team to really keep an eye on.

What are the leaks and how do I fill them? And we're really excited about this feature because it's going to make that process of filling those leaks feel a lot more ergonomic and a lot more impactful. And you can say, Hey, I've just improved the quality of 30 people's conversations with an agent suggested response and a click of a button, which is pretty cool.

context isn't one thing. It's a bunch of different things and the levers and the buttons, depending on when you push them and at what velocity and at what pressure and when and all of that matters and like tuning the agent's output. Because as we were saying before, know, like agents are getting smarter, they're getting a lot better, no doubt, but they are not perfect. They are not infallible and they need a lot of direction. And that direction is

the data teams, I think responsibility going forward is they, they need to really take ownership of saying, Hey, this is going to be the best agent. And it's going to be because we put the right context where it's needed. We followed these workflows. We kept an eye on things. We measured success and we measured the way that people are using it, what questions they're asking. And now we have like a new signal, even for ourselves to say, Hey,

We had five people ask this really interesting question and the agent got it wrong 100 % of the time and they had no context to figure it out. And in fact, we don't even know how we would answer that. So that's a great thing for us to go take a deep dive in and now we're not sort of stuck in the mud, but we also have a way to sort of resolve that in the future because we're gonna do the deep dive that we would have done. We're gonna figure out the right answer. We're gonna canonize all of those metrics and the learning inside of HEX.

and they're gonna give all that back to the agent so that it gets smarter. And that just feels like the nirvana that I think we've all sort of been reaching for for a long time.

Ben Jones: Right, I mean this is kind of the notion of continuous improvement, right? And making knowledge institutional as opposed to something that just floats off into the ether and disappears with a given analysis or project or what have you. So it's the ability to take what was learned and incorporate that and share that and spread it and then be able to actually improve upon that again next month and next month after that.

you know, giving organizations the tools to continue to improve, I think is key. And so, you know, it seems like you're turning the data team into just continual change agents. So I want to know a little bit more about that, like on the inside at Hex, a lot of people, course, following your company, your products, thinking very highly of it. What's it like for the data team within Hex? I'm assuming they use the product quite a bit and look, what does their role look like? What, what sort of

influence do they have within the organization when it comes to helping inform leadership, actually making decisions, and even improving the product itself.

Rachel Herrera: Yeah, tons, which is fantastic. And our data team are probably like the MVPs in a lot of ways because they have allowed Hex to be the tool of choice for, think it's the metric I saw recently, which is just crazy. Like 80 % of our company uses Hex every week. So that's salespeople, leadership, marketing, design, engineering, operations.

using it to do all of the things that we are talking about. And the only reason they can do that is because the data team invested and believed that they could curate data, create the semantic models, train the agents, write the guides, do all of the foundational work that's required. And then the adoption just exploded, which is fantastic. And so now there are days, and I actually sat next to

a member of our data team, a woman named Beth. She's in New York and she's fantastic. She lets me bother her an hour a week because I'm just keenly curious, what does her day look like? And I also just want to stay close, like ear to the ground. You what are you experiencing? What are you struggling with? And she's building less dashboards, but she's building, the dashboards she does build are like incredibly important. Like they are highest priority, need to have this done, highest level visibility and

maximum business impact, which I think is really cool. They're not the, I have a kind of quick question, it sort of morphs into a dashboard or data app, and then it sucked a few hours of her day. She's spending a lot of time looking at threads and how agents are responding to users' questions, where the gaps in the context are, but also giving feedback to our product team. Hey, the agent said this, this was weird.

where the agent said this, and this was really insightful, we should do this more often. Or hey, this part of the workflow doesn't feel quite right. So they're almost like the ultimate alpha testers of the product, which is great. And they're answering a ton less quick questions, Slack, DMs, that is almost gone. And not to zero, because I think that'll always be there, but they're spending so much more of their time improving data quality, data pipelines.

cleaning the data, pulling in data that has been difficult to get incorporated into our core data sets. They just haven't had time for, now they have it. And now the agent has access to that data, which is super cool. So the days are not short. They're not twiddling their thumbs by any means, but they're doing things that are kind of more in the vein of what I'm expecting a lot more data teams to look like, where they're less time building repeatable things and more times curating stuff that scales.

a lot more easily and then spending the time on those deep dives where it truly like, I couldn't get this answer. And that's the other cool thing that I think I've heard is when people come to them with a question now, it's not, hey, here's my question. It's, hey, here's my question. Here's how I tried to solve it in threads. Here's what the agent gave me. Here's why I didn't think that was right. So I asked it these other questions and here's where I got stuck. And so now the data team has this like ledger.

of what this person was trying to do. And they can get so much faster, like closer to, ⁓ I see, versus the, like the turns that I was mentioning before where you're, okay, not quite right. Okay, not quite right. No, not, that's not exactly what I was looking for. And that's the part that I've heard has been like the coolest and also just seeing it. And I'm like, man, I'm so jealous. wish when I was on data teams, I had that.

Ben Jones: Same. mean, those are richer conversations. Those conversations have a lot more information within them. You know, in the past, it was really just, well, how many people are viewing this dashboard? You had really beyond that little idea what their experience was like, what questions they actually asked and answered, and if they felt the experience was helpful or not, if they got

or whatnot, you just really had no idea. So it's not just a question of like, you you mentioned the 80 % number.

It's so impressive to think about that, but then also to go beyond that and to say, it isn't just that we know they're using the tool, it's that we know how they're using

we can do to improve the outcomes they get when they use it. So it isn't just a question of like adherence or utilization or let's say quantity of usage, but actually additionally more around the quality of the interaction and how well it went and what we can do to improve the quality.

and the quantity at the same time, think that's where you start to empower those data teams to really change the game within a business.

Rachel Herrera: Yeah. Yeah. And I think it also like speaking to the future state of all this too. One thing that we're trying to understand more deeply and get right. And I think a lot of other data team members are as well is like agent evaluations and how do you even automate some of that work that maybe now require somebody to ask a question to get stuck? Could you. Precede 50 questions, give those to the agent and understand how it would respond before a human even has to touch it. And so.

That I think will become another really important core competency in this sort of curation workflow that I've described here, which the more I talk about it, the more I'm like, that is a full-time job. Like that, that requires somebody to be on, on task, paying close attention, being very close to the business. And that's the stuff that I love to see.

Ben Jones: Well, they're certainly going to be a lot more influential if they can pull that off. But now here's the next question. You know, as I mentioned, I teach in the university and people are going through these analytics programs that, you know, may have caught up, may not. My question to you is what do you think analysts should do, whether they're in school and learning to become one or they already are one? What do they need to do to prepare themselves for this change to curation and context and decision support?

⁓ How can they get ready for that, both in terms of their skill and their knowledge, positioning themselves and understanding how they can evolve into that potentially new, more influential role within organizations?

Rachel Herrera: Mm-hmm. Well, I think the basics and the fundamentals stay the same. Understand SQL, understand basic analytics, best practices, understand when to use what sort of methodologies. Like all of that stays the same because you need those to just be able to audit the agent and make sure that it's doing what you expect, but also to teach it. Because if you don't understand the fundamentals, how can you teach an agent to sort of think and reason the same way that you would expect somebody to?

But think the other important thing is, like I said, it's the soft skills part. It's sit next to a business user and be comfortable asking them about what they care about. What did they do? And also how the business itself works. How do they make money? How do they lose money? How do customers interact with them? Because the closer you can understand those things, the better you're going to be able to translate the insights or at least shape the data in such a way that's going to let that person make best use of it and make an agent be useful to them.

If you are very confident in your technical abilities, that's great. Now beef up the, have great discovery skills, have great sort of like, okay, ask why six times to really understand. And like all of those little kind of tricks that we used to use to get to a good analysis that we'd have to go then handwrite, those all still very much apply now, but I think you have to do them a lot sooner because before you sort of had like a year or two buffer of being a junior, being able to just heads down.

kind of right against your keyboard and then you finally were able to go outside in the big world and like talk to other people versus just take tickets. I think that's going to happen a lot sooner now than it probably did in the past. And so being able to have those skills early and often is going to be really, really critical. And that extends also to storytelling, know, being really good at finding an insight, but being able to articulate why it matters and the decisions you should make from it.

so in making that compelling to the business, but not making it a 15 page slide deck, it's two sentences. It's one metric. It's one chart. You know, how do you edit yourself down so that you tell the most important part of that story? And that's, think a lot more challenging than people realize, especially when you care about what you're doing and you know, it is all important. I'm even guilty of it. I'm sure in this podcast, I've used 10 words when five will do, but you know, I'm a data person and I need.

Rachel Herrera: Nuance is a dirty word, so I need to be explicit.

Ben Jones: Guilty is charged as well.

Yeah, I love my editors. That editing process is painful. Data is like that. We come with all this data. We want to show it all so that we can get that convincing argument. But at the end of the day, you're right. It's kind of coming down to, do you understand your audience? Do you understand them deeply enough to be able to articulate in the clearest, most compelling way what it is that they need to know?

Rachel Herrera: It is.

Ben Jones: And that's the human element, that's in the human realm. And that's where we need to lean into that and focus on listening, being curious. And I agree with you wholeheartedly. I think that's actually a really welcome change. I would rather do that than push pixels around a screen. I mean, I don't know about you. So maybe this could be a good thing. I think there's so much fear out there. One of my future episodes is gonna really focus on just that, this question of fear and how we're all kind of processing what's going on.

But maybe there's a hopeful vision as well of this being a new future where people who are specialists in data can just kind of become a lot more influential and ⁓ be involved in the human elements in the realm of the business itself and understanding business users and customers and just really making things better in that way. So I'm hopeful that that's where we're headed to. And again, I'm really kind of a big fan of what you all are doing at Hex. And so, Rachel, thank you for writing that article. I sent it to all my students.

the idea that you need to be a great storyteller that really resonated with them. And I think that they were relieved to know that people in the industry see a way forward. And so looking forward to reading a lot more of what you write. Looking forward to this new launch, you kind of hinted at a little bit with context studio agents being able to help us review and provide recommendations and suggestions to improve our systems going forward and just build that.

virtuous cycle of continuous improvement of the data and the answers we're getting therefore So I would just want to thank you again for being on the show here and wishing you all the best and success here in the next little while as you continue to launch brand new features at Hex. And anything you want to wrap up with on your end?

Rachel Herrera: Definitely check us out. We let you do all kinds of other stuff that I didn't even get a chance to show, like build notebooks, build charts. Just have a ton of fun. I think that's my favorite part about Hex is just playing around writing and data and getting back to some of the fun that maybe was missed in the past, but doing it all in one place is also my other favorite part. So definitely just get your hands on it, try it out, and don't fret.

Agents are cool. can teach them

way, you know, and we can make them an extension of ourselves rather than feeling like they're this sort of foreign body that we have to kind of fight against or like compete with. I think that's helpful framing for me, especially is realizing that an agent is only as good as the person that's trained it in a lot of ways. And so if you're an exceptional analyst and you teach an agent,

Ben Jones: Mm-hmm.

Right. Mm-hmm. Mm-hmm.

Rachel Herrera: they're going to be an exceptional agent and that just multiplies your value 10 times over. And so maybe that is what I'll leave us with is just remember that they're here to work with us,

work for us, and we should own that.

Ben Jones: Yep.

I love that. Yep. Have fun with it, play with it and find a way to use them to expand and extend your scope and your influence. That's really what it's all about. So you again, Rachel. Wishing you all the best. You take care. Okay. Okay. Bye now.

Rachel Herrera: Thanks, Ben.