Building and Validating Research Agents in GTM

Julia Nimchinski:
Wow, Patrick, welcome back! Thank you very much, Julia. Nice to see you. Yeah, nice to see you. Long time no see.

Patrick Spychalski:
In a minute.

Julia Nimchinski:
Congratulations.

Patrick Spychalski:
Oh.

Julia Nimchinski:
Thanks, much, I appreciate it. What’s the latest and greatest?

Patrick Spychalski:
You know, just, as, you know, as you congratulated me before, a company got acquired, we’re still working at it, and, building a lot of cool stuff, and, pumped to show off a few use cases today.

Julia Nimchinski:
Let’s dive into it.

Patrick Spychalski:
Awesome. So, yeah, the presentation I wanted to do today, and for those who don’t know me, my name’s Patrick Spotchowski, I run an agency that, called The Kiln that implements, AI go-to-market tools for enterprise companies, and The general topic I wanted to talk about today was just a few ways for people who are new to agents and go-to-market to implement these tools within their businesses easily. I just feel like there are a lot of, you know, LinkedIn posts, Reddit posts, Twitter posts out there, where people are showing off some of the agents that they’ve built for their companies, and often they’re very complex. It requires a lot of setup. And… and a lot of, you know, technical acumen, and I feel like it can get somewhat confusing, and I… and then there’s also so many different types of agents now, like, there are… there are research agents, there are agents you can connect tools to, there are coding agents, right? So there’s all these different things that you can… you can be prompting and using, and I just wanted to kind of give, like, a… a demo on the things that our enterprise clients seem to be using and enjoying the most, so I figured I would just show off some basic ways to build deploy agents, a couple tools that I liked, and yeah, that’s the plan.

Julia Nimchinski:
Very cool.

Patrick Spychalski:
That’s it. Awesome. Sounds good. I will share my screen. So the first tool I wanted to show off is Clay. You know, for full transparency, our agency probably implements Clay more than any other tool out there. And I think Clay has particularly good research agents for go-to-market. So, you know, agents that can go and extract publicly available and sometimes privately available information. And then sequence it into kind of an easy-to-use or easy-to-read format. And so Clay has their own agent that they have you know, aptly named ClayGent, and I wanted to kind of show off how this agent’s usually set up in the tool, and how you can use it for your business. And so, just as a quick rundown, for those who don’t know what Clay is, it’s more than just an agent tool, it’s kind of this… The best way to describe this is maybe a really good workflow tool for go-to-market, combined with a really good enrichment tool for go-to-markets. I think, like, Zapier combined with, like, You know, a better version of, you know, your… whatever enrichment tool you’re using. And one of their really great enrichments is Collagent, which you can prompt to find information from you for the web. So, I wanted to show off a very basic example of this in a table that I’ve built in Clay. As background, this table is just enriching CRM leads that I have. And you can see it’s very easy to set up. The first thing you do is, just choose the model that you want to use. In this case, I had GPT-5 set up, but, you know, pretty much every OpenAI model you can think of is here. And even if you don’t want to, you know, even if you don’t see the model you want to use here, you can connect your own OpenAI account using an API key, and connect to the agent. So next, this is a very basic prompt. For full transparency, I would make a much better prompt than this if I was deploying this, like, truly within my business, but the idea is just… in this case, I wanted it to do all of the research it possibly could about this given company and my CRM’s internal sales team. So I said, scrape your, you know, website, review sites, case studies, literally whatever you need to do, here’s their website, go. And I think it pretty aptly shows how good these agents have gotten. When I first started using Collegiate, it was kind of in the early days of the OpenAI APIs, and sometimes it would hallucinate really heavily, but now it’s pretty much only returning super accurate data, and I think it’s doing so in a much better format. Like, you can see there’s a confidence score for the information, it shows you all the different steps taken, it shows you how many, how long it took to actually return the information, in this case, a pretty long time. And it just gives me this really massive rundown, in this case, of Brex’s internal sales team. It goes through, what their sales team members are getting paid, the different structure of their sales team, what tools that they’re actively using that they could find online, and it’s scraping pretty much anything in the web. It’s scraping, you know, like, obviously Brex’s website, their LinkedIn profiles, the LinkedIn profiles of their leaders, their job listings, and it’s outputting all this information. And of course, in this case, it’s a really long block of text, so it’s kind of like, what do you do with this, I guess? But there are also ways you can structure these. So, like, in this case, I said, give me the revenue model of this business, and it was able to output the model in kind of just a comma-separated format. In this case, I prompted it to give me some use cases in clay. And it was able to output in a bulleted list. So, with a lot of interesting stuff you can do with clay agents. Again, I think it’s best for research, but you also can connect tools now. You can see at the bottom of of the… maybe it’s this one? You can connect MCPs now to Clay’s agents, you know, and I would say the performance is not, like. as good as probably, like, the, you know, other tools I’m gonna show you in this list, but you can also take action. So, like, in this case. I actually had this agent build me another clay table. Like, I had it actively log into my Clay account, and then build a clay table using a clay agent, so it’s kind of like Clayception. So, yeah. This is the… kind of first example of a way to use agents in go-to-market. I think it’s a great way to just find and research information that maybe isn’t as structured or can’t be provided by a conventional data provider. Highly recommend using it, and so that’s the first one. I will move on to the second, unless you have any objections, Julia. I’m guessing not. Okay, great.

Julia Nimchinski:
when you showcase these, the only objection is, how, like, how would a GTM leader actually, you know, deploy this without a technical skill?

Patrick Spychalski:
Totally. So, I mean, what’s great about Clay is it is a no-code tool, and relatively straightforward. I would say, like, while there is a bit of a learning curve, you know, if you’re… like, a head of RevOps, or, you know, even just somebody in the RevOps org, or even in the sales org, I think you’ll get used to it pretty quickly. So, like, I’ll just go quickly back into my Clay account here. And let’s just say I’m creating a new workbook. all of this is pretty intuitive, like, you can just find a list of people. It has kind of, like, ZoomInfo-like filters, so, like, you know, let’s say, hypothetically, I wanted to find somebody who, you know, is a CEO, and, you know, they’re in New York. Sure, we’ll just make it, like, pretty simple. Not Anguilla, New York. So, you can limit this to maybe, whatever, 100 people. Maybe we’ll even do, like, just 10 to keep it simple. And so we have 10 people that are CEOs based in New York now, and you can import them into a table. Like, again, none of this, I would say, requires any sort of technical acumen. And then, you know, let’s say you wanted to run an agent. It’s… you just type in Collagent, and you’re able to find the agent. And you can prompt it for whatever information you want, as I showed you, like, it just pops out a column and it gives you some information. And then, like, you know, whatever, wherever you want to send this data, it still doesn’t take a ton of technical ability. Like, let’s say you wanted to send it to, like, Salesforce. And you have all these leads, you want to update them to Salesforce. you just click on, maybe, like, a Create Record integration in Clay, you can connect to Salesforce, and you can very easily just pick whatever records you want. In this case, it would be a contact. And then you just map the columns to your Salesforce. So, like, you can just throw in a column here, first name gets mapped to first name. And then when you’re done, you just run it. So, it isn’t like, you know, I would say a super technical setup. I would actually say this is one of the I tried to show, like, different levels of difficulty for these agents. This is probably… I have kind of, like, four demonstrations. This is probably the second most difficult to set up out of the four. So, like, you know, I try to keep it pretty simple. I figured, you know, a lot of the audience are, you know, sales leaders, people who aren’t engineers that want to build agents in their work, so that’s kind of the approach.

Julia Nimchinski:
Cool.

Patrick Spychalski:
Great. Let’s move on. Cool. So the second is N8N. This is probably the most complicated way that I’m planning to demonstrate setting up agents. N8N is a much more flexible workflow tool than really any of those out there. I would say, in many cases, it’s kind of a developer-forward or developer-first workflow tool, but because of that, it allows for a ton of flexibility. And so I wanted to show off a a very, very basic N8N agent. And if you’re wondering how this workflow is built here, it’s actually quite simple. You have a trigger, which we have set up here for when a chat message is received. I think this is coming from Slack. And then, we have this agent, and you can see, if you type in agent. it should pop up. AI Agent pops up right here, and you can start creating your agent. So I’ll show you how this one is set up already. Delete this one that I just made. Essentially how these agents work is you have a, like, a user message prompt here. You can select the type of agent you have. In this case, we’re connecting our agent to a bunch of different tools, but you can see there’s, like, a SQL agent, an execution agent, a conversational agent. In this case, we’re connecting ours to a bunch of tools. And the idea with this workflow was pretty simple, alright, you know, whatever you want to call this. It’s like a campaign chatbot, so we have a lot of these campaigns we’re running in an outbound tool called SmartLead. And Smartly doesn’t have an internal tool to allow us to output analytics based on a natural language prompt. So, for example, you might be asking. hey, I’m running 15 different email campaigns, which one of them’s performing the best, which one’s performing the worst? And then, you know, give me some copy analyses on both, and tell me some commonalities between, you know, the successful campaigns versus the non-successful campaigns. That’s a pretty common question, I think, especially in a lot of sales teams. You can’t really ask that to your tool, at least until now. So, we were able to build an MCP for SmartLead. For those who, you know, haven’t used an MCP before, it’s just essentially a way to talk to a tool. Using their API. And we’re able to connect these to our agent. So we can say, you know, in the instructions, we want to be able to answer questions about our Smart Lead campaigns, and you can see here it’s using, you know, a Slack MCP and a few other MCPs to just extrapolate data from our clients. As well as the, the Smart Lead campaigns, and then it will return you an answer based on a question. So, you know, if we asked it, you know, again, that good versus bad campaign question, it would return that information right back into Slack. And create an agent that way. So, this is kind of a step up from what I just showed. The first thing I showed, in Clay was more of a research agent going and browsing the web and returning data to you. This is a little different, where you’re actually connecting existing tools that you’re logged into to an agent to give you more first-party data. So. It’s… it’s allowing you to use, kind of, internal business data to respond to a message. So, that was the next one, and I think a pretty… a pretty clear, maybe, rebuttal to building this is, like, again, this doesn’t look that complicated, but when you build a really, like, can-be-deployed-in-your-company agent, it comes with a lot of complications. Like, for example, you know, maybe you have like, you know, 40 BDRs, 20 AEs, and then, you know, some leadership, they all probably should have different permissions to this tool, right? Like, maybe you only want BDRs seeing certain campaigns, or maybe you don’t even want them to be able to ask questions to your CRM. And then it’s like, well, what CRM instance do you connect? Do you connect your connection? Do you connect your BDR’s connection to the CRM? So there becomes… there’s just a lot of administrative problems associated with building these agents that take a lot of technical work to solve, and so that’s where I wanted to show this next tool called Dust. So I think Dust is a really interesting tool. It’s… pretty much a way to deploy agents on a larger company scale. So, for example, Clay, one of the tools I just showed you, they’re a really big company now, like, maybe not really big, but, like, 300 people, you know, $5 billion valuation, they have a lot of people on their team. And they actually use Dust to, as an internal knowledge base. So they have a bunch of different agents in Dust, and they can ask it questions within their org, and Dust will respond. So, what I really like about Dust is you can create different agents for different tasks, and those agents can be connected to different tools. So. I created a really unoriginal name here called Mr. Agent, and what I’ve done with this agent is I just wanted it to be a general knowledge agent. I said it should be able to answer general questions that I ask about our business. And I’ve connected, as you can see here. my Google Drive, my Notion, my Slack, my HubSpot, my Gmail, and you can connect really any tool that you’re actively using. Like, if I was to do this again, I would create… I would connect Monday.com, I would connect, probably, my other couple emails that I have for our business, like, pretty much anything I could possibly think of that would have knowledge about my agency. And what’s really great about it is, A, it’s super easy to set up, and B, you can also change permissions for these agents. So, like, if you go into, the administrative panel. You can see that there are… Different members here, with different, access permissions. And then, in the workspace, when you’re connecting tools, let me see if I can find the tools section here. They’ve moved it around since I last used it. This is tragic, can I even find it? Maybe not. Either way, when you connect tools, you can, either make them public or private for your… for your company. So, like. I don’t really want people asking questions to my Gmail. You know, my Gmail would probably be private, and so you can set it on private, but then maybe you have a company-wide HubSpot connection that you can make. It’s very easy to make that, very easy to make the HubSpot connection public. And so, I created this new account, I made this new agent to just demonstrate this, but I can see, like, okay, like, what have we been doing for Windsurf, which is one of our clients? And it was able to give me this high-level rundown using Slack messages, emails, our Google Drive, our internal Monday, and it was able to respond and give me all this information about what we’re working on with our clients. So the high-level reason I recommend checking out Dust, and our team started to deploy it within their, within, you know, their workflows as well. is that it’s really easy to set up. Like, it took me, like, 10 minutes to set dust up. The agents are really well made. They’re actually not using your conventional chat models. They’ve created their own models specifically meant for tasks like this. And it has a lot of administrative features that I think are great. So, highly recommend checking out Dust. It’s been a great tool, and, the Clay team is constantly using it in really creative ways. So yeah, that’s the third. And then the last one, I’m actually not even gonna demo, because I haven’t used it enough to put an opinion on, but this is ju- I just wanted to talk about Where we’re planning to go, which is… we… our whole team uses Claude a lot. And they just recently launched a… pretty much, like, a dust competitor. In Claude, so for anybody who uses, you know, Anthropic products a lot, like, our team is all, like, we’re all in this big Claude Code team plan, and, like, all of our team uses it for a bunch of different things, and they pretty much just set up, created this new one, so I just created, like, an example account to show you this, but you’re now able to connect pretty much the same things that you’re able to connect to Dust to Claude. And then, you know, add a bunch of different permissions, etc. You can see, you can customize the different permissions in Claude. And then, of course, set up your own internal bots within the Anthropic product. So, if you’re not looking to acquire any more tech, while I still think Dust is probably a little bit more built out and mature. Yeah, we’re probably gonna move pretty quickly to Anthropic, just because we’re obsessed with cloud code, and it’s integrated directly within it. So, yeah, those are kind of the four different ways, we’ve been using agents. we’re seeing our clients using agents, and as you can see, not a lot of them are super, you know, flashy. It’s not like, hey, we built this agent to replace, like, I don’t know, a company’s head of sales and, like, run everything, but we try to, you know, as an agency, build what’s practical and, like, immediately value-adding. I think that’s what’s great about professional services, is, like. when companies hire you, you actually have to build something that adds value, or you just get fired, and so we kind of get to discern whether something on LinkedIn is just fluff that doesn’t actually help anybody, or genuinely can be used by all, can be deployed easily, and these are some of the tools that we’ve seen deployed well in all these different companies.

Julia Nimchinski:
Super valuable. Thank you so much, Patrick. I’m just curious, like, touching on this last point you made. I still work with so many clients, In terms of revenue leaders, specifically, like, mid-market, enterprise. We are trying to be really careful when we talk about AI ROI. In the community. I mean, more specifically, we don’t talk about it. But since you mentioned, what are you seeing? Is it too early to see any meaningful ROI? Is it more like, I don’t know, just productivity improvements?

Patrick Spychalski:
Yeah, so, actually, you’re kind of in the second half of your question and answered it, which is most of AI implementation has been efficiency gain, where value’s actually being created. So, for example, something a lot of our clients will build are these research agents connected to something that sequences it into something a little bit more readable. So, like, whatever you have 100 BDRs on your team, they’re spending 2-3 hours a day researching prospects, and in reality, you can prompt even a, you know, decent model agent to replace this research for you. And then, of course, effectively save them time. And so, I think that’s where we’ve seen a lot of the… efficiency gains. As you can imagine, we work with a lot of enterprise companies, and they’re not the fastest-moving companies in many cases, and so we really have to be like… but they want AI to be implemented. Their boards are pressuring them to implement AI, the C-suites are pressuring them to implement AI, so it’s like… What is pretty much the most safe possible use case to implement? Across, like, an enterprise org, and a lot of it is, like, data enrichment using research agents, or creating efficiencies in sales teams by automating research. But, you know, a lot of these enterprise businesses are not using, like, the, you know, whatever, AI SDR tools, or, like, completely, you know. replacing their top of funnel with AI, like, none of that is really happening, so… when we’re pitching a client, a lot of it is… your team members are going to be saving time, because this very low-level task can easily be, you know, replaced by AI, and it’ll allow them to do more, you know, let’s say high ROI things, like whatever, cold calling for a BDR, for example.

Julia Nimchinski:
Chris, you mentioned your, natural inclination towards, you know, the highest ROI friendly, like, even in terms of acceleration and productivity enhancement, indicators, just… That sounds like. That’s specifically the tools that you’re focused on, but curious, what’s the process internally, like. you work with clay, Anthropic, perhaps, some others, but Do you even plan to expand into all of the AI SDRs or some other tools? Like, what’s the philosophy there?

Patrick Spychalski:
So, the short answer is, like, a hard no, actually. Like, we… I really don’t believe in many cases that these tools, at least in the short to medium term, are going to be able to replace SDRs, and frankly, the revenue leaders we work with at enterprise companies have no interest in doing that. Like, to this day, not once have I gotten on a call with, like, a CRO at a large company, and they’ve said, we, like, we want to replace our SDRs with AI, it’s always… we want to create efficiency gains with AI, and . You know, so the tools that promise to cover the end-to-end role of a sales team member, in our opinion, haven’t been super effective. And frankly, tools like Clay allow for, you know, the nuanced strategy of a skilled CRO to just be realized at a higher scale, whereas we don’t want the agents or whatever models we’re running to be doing the strategy for us, which is, in many cases, what these agents are promising, so… I would say, yeah, and not anytime soon. I think tools like, you know, Clay and N8N and Claw are ways to just very quickly deploy systems that are run by a person doing strategy.

Julia Nimchinski:
And in terms of your intake, how would it look like? How long does it take to even start working with you?

Patrick Spychalski:
It takes quite a bit of time, I mean, mostly because of discovery. Another thing, a misconception that I see people kind of fall into when implementing AI tools is it’s so tempting to go out of the box and just, you know, use the tool that you can set up in 15 minutes and just click a button and run. But the actual, you know, ROI-generating systems usually take a lot of scoping and custom building. Every enterprise is different, they have different CRM setups, they have different ICPs, they have different ways of doing outreach, and their sales orgs are structured differently. And so when we’re deploying systems, we want to make sure that, A, they can be adopted by people very easily, and then B, they actually perform the task that they’re set up to perform. And so, you know, we have for a lot of our clients, quite a few discovery calls, figuring out, you know, whether it’s even viable to build a system for them, and then if it is, the best way to go about doing it. So, I would just say, for anybody who’s looking to, you know, build complex internal AI systems to save time, I highly recommend you know, kind of going the more technical and complicated route, because it’s often, maybe not complicated, but the more technical, bespoke route, it’s usually a lot more effective than just clicking, you know, play on a tool.

Julia Nimchinski:
Definitely. And in terms of just, you know, failures and… and just… because we typically focus on, you know, the most successful scenarios, especially here, but based on what you’ve seen, like, why would, you know, certain clients succeed just working with you, and the other might be having some complications?

Patrick Spychalski:
Yeah, totally. I mean, so it depends on what the… the thing we’re building is. I mean, for example, one thing that I think a lot of revenue leaders are looking to outsource is, you know, outbound campaigns using AI tools, and of course, most people look directly to, like, the copy, and, like, generating the copy with AI is, like, the core strategy for efficiency gains, but, like, we’ll talk to a company, and they’ll have, like, the worst email infrastructure ever, and they’re getting, like, 10% open rates, and it’s like, well, that’s… your problem isn’t really the copy, actually, you’re just having 10% of the people you’re sending to open your emails, so… I’d say, ultimately, a lot of failure is infrastructure failure. Another great example is if we have a client, with really, really bad CRM data, and they want to build something kind of fancy, but the basis of that system is on their CRM data, and then it’s like. Well, your data’s so bad that it’s really hard to, like, you know, build a research function using this data, because we don’t even have the person’s email or their LinkedIn profile, and that’s kind of the basis for the research. So, I’d say a bad data layer, or, yeah, a bad data base layer is another thing that we see pretty often. And then also, the last one I’d say is just… you know, and maybe… I don’t know how to put it, like, kind of, like, lazy hiring, like, where you just bring on an agency, and then you don’t really… like, you just say, like, go build stuff, and you don’t really engage with them. Like, ultimately. even the best agencies in our space, I still think have worse strategy, probably, than, like, a CRO about their own company, so I think it’s really important for there to be a real revenue leader in charge of the systems that are being built, and then, you know, outsource the deployment to somebody like us, or maybe even hire somebody in-house to do the actual implementation work. But, you know, you’re still gonna know your business better than any agency you hire, so I think it’s really worth thinking deeply about what you want built, and whether it’s actually worthwhile to build before hiring somebody and just kind of, like, deploying AI willy-nilly.

Julia Nimchinski:
Yeah, totally. In terms of a good email infrastructure, could you just share more on your thinking here? What does it look like?

Patrick Spychalski:
Yeah, so, cold email, as I’m sure everybody here knows, is getting harder and harder to, to run at scale, and it takes, actually, quite a bit of, let’s say, focused work to build good infrastructure, so… You know, it first involves building, you know, alias domains. You should never use your core domain for cold email outreach. And then, you know, it’s also then the setup of said domains and the email accounts connected to them, so we usually do 50% Gmails and then 50% Microsoft accounts, and we’re actually setting up all these accounts manually, because most of the tools that promise to do it automatically have shortcomings, and I can talk about those in just a second. And so we’re setting them all up manually, which is a pretty arduous task. We have a guy on our team whose full-time job is managing our email infrastructure for our clients. It is a really, you know, it is a full-time job, especially if you have, like, you know, 30s on the clients. And then once they’re set up, there are limits to what you should be doing with them. You should be warming them up for 2 weeks, you should be sending, like, 10 to 15 emails a day per account to make sure deliverability stays sound. And you should never be using private servers unless you actually own the server, so there’s just a lot, kind of, that goes into email infrastructure. And I think a lot of companies, again, just kind of connect a bunch of rep accounts to a tool and start blasting emails, and then it immediately crushes their deliverability. And then, in many cases, the tools won’t have open rates turned on. So then. the CRO’s like, why aren’t we getting responses? And they can’t see their open rate, so they’re like, oh, well, the copy’s probably bad, and then you run a couple checks on their accounts, and it’s like, you know, the emails have been trashed for 6 months, and they’ve been sending emails to nobody, and… So yeah, it can just be… I honestly think infrastructure’s, like, probably the problem for 50% of the companies we talk to. It’s not even, like, strategy, it’s just, like, they have bad email setup.

Julia Nimchinski:
So when I start working with you, you will just essentially rebuild the whole infrastructure from the ground up.

Patrick Spychalski:
Yeah, we used to try to salvage our clients’ infrastructure, like, we would try to, like, okay, we’ll rewarm the emails up, and we’ll try to use the same domains, and we’ve actually completely stopped doing that. Like, it just doesn’t… like, it’s just a lot easier to build it from scratch. So, you know, even if you don’t hire us, I would highly recommend, if you even have an inkling that your emails are not set up correctly. Hiring an email infrastructure expert and getting it built properly, because it’ll solve you a lot of headache.

Julia Nimchinski:
Patrick, you’re advising a lot not to hire you, but… Let’s reverse this. Why would they hire you?

Patrick Spychalski:
I mean, so the sort of work we do is classified under this term, GTM engineering. I’m sure some of you have heard about it, it’s kind of like a growing job title. And it’s pretty much somebody who can strategize and then deploy go-to-market systems using newer AI tools. And, I think it’s a pretty hard job when there’s not many of them out there currently. Like, you know, the demand for GTM engineers are much higher than the supply. And I also just think that an agency, separately, is kind of this, like, collective aggregate of knowledge on a thing. You know, you hire a marketing agency, you’re hiring, whatever, 30, 50, 100 people who all know marketing really well, sharing their knowledge collectively to build something for you. And so, we kind of feel like we’re that for go-to-market engineering. You know, you could hire somebody internally, and they’ll be actually significantly cheaper than us in many cases. But you’re getting one person who knows how to build one sort of system. We’ve worked with some of the largest, you know, tech companies in the world in building these systems, so we know what’s right, we know what’s wrong, we know what adds value, we know what works, we know what doesn’t work. And we also just have, you know, a lot of, kind of, manpower to deploy to build these systems faster than somebody you could hire internally, immediately, so… you know, that’s usually our pitch to people, but also, you know, we’re having… I just said go-to-market engineers are hard to hire, so we’re honestly throttled and constrained by the amount of people we have on our team. So, you know, I’m not gonna ever push people against hiring us, but, you know, it makes it… makes it tough to bring on new clients when it’s hard to hire people. And Jordan just joined. Sweet. What’s up, Jordan?

Jordan Crawford:
Heather?

Julia Nimchinski:
Always great featuring you. Patrick, what’s… what’s next for 2026, and where should our community go?

Patrick Spychalski:
We’re still planning to just build things and clay any den in Cloud code, that’s kind of our plan. So, you know, obviously, you know, feel free to give all three of those tools a try, I highly recommend them. And… I’m pumped that Jordan’s after me, because he’s also a very early user of all three of those tools, and incredibly good. So, yeah, appreciate it.

Julia Nimchinski:
Super cool. Thank you so much, and we are transitioning to our next session.