The AI Operator Podcast: Why Outbound Tools Break Your CRM

Most sales teams don’t realize how messy their CRM is—until it’s too late. In this episode of The AI Operator, Jordan Park interviews OutboundSync founder Harris Kenny to talk about the hidden cost of bad integrations and disconnected outbound tools.

From his early days running a fractional sales agency to building a product that bridges Smartlead, HubSpot, and Salesforce, Harris shares what he’s learned about fixing data chaos, scaling outbound the right way, and avoiding the trend-chasing trap.

If your CRM is a black box or your workflows feel like duct tape, you’ll want to listen.

Watch

Listen

Transcript

00:00 What is up everyone? Today on the AI operator, we are joined by Harris, Kenny, the visionary, ah, behind Outbound Sync platform revolutionizing out B2B sales teams integrate their Outbound efforts with CRM systems like HubSpot and Salesforce.
00:16 He’s got over 15 years in sales and go to market strategies, ah, Harris identified a critical gap, the disconnect between modern Outbound tools and traditional CRM systems.
00:27 So Outbound Sync addresses this by seamlessly syncing outbound campaign data, reducing manual tasks, and ultimately enhancing data accuracy. Now, prior to outbound sync, Harris founded intro CRM.
00:39 It was a fractional sales agency and HubSpot solutions partner, where he really honed in his expert in sales operations. So we’ll be looking at his journey from consulting to SaaS entrepreneurship and the valuable insights into a valuable ever-evolving landscape of sales technology.
00:56 So with that, Harris, welcome to the show, man, we’re super excited to have you. I’m excited to dive in because I were a similar breed, right?
01:07 HubSpot Agency, closed agencies, automation and workflow process, sales process, heavy, which not everyone on this show fits into that box.
01:16 And then you also have the extra edge of being, you know, a SaaS founder. So super excited for this conversation.
01:23 How are you doing today? Doing great. I really appreciate the opportunity. And yeah, I mean big fan of your stuff and And Mariah and your team, I mean, I’ve seen your stuff on LinkedIn for years.
01:32 So yeah, I’m excited to chat and I’m excited to talk about what’s new on the AI side of things too and how this all kind of comes together because there’s a lot going on today.
01:40 Just like the day that we’re recording this, a ton of things have come out. So it feels like there’s never a moment right now.
01:45 There is not. It’s hard to even keep up with now as part of the intention behind this podcast was to drive a little bit of awareness of where people might be falling a little behind, you know, where people are and ultimately like how we see this evolving because you know that my best guess is is you 
02:05 know I mean it’s endless like we really don’t know at this point so we’ll talk a little bit about that but I do want to start with your background as an entrepreneur as an agency owner and really the transition that you kind of have made so if you would can you share a little bit about your journey of
02:25 like the founding intro CRM and then ultimately how you ended up launching an outbound sync and what inspired that transition.
02:34 For sure. So I had worked in a variety of sales and marketing roles prior to starting my agency. And I actually had a lot of experience in hardware.
02:44 I scaled a 3D printer company from yeah, from like 1.7 million to 20 million in annual revenue you and scaling hardware company is exceptionally difficult.
02:55 Just the challenge of having to buy things ahead of time and having the labor and things like that. We were manufacturing the 3D printers in Colorado, you know, sort of before it was cool, before all the latest trade conversation.
03:08 And so we scaled and learned a lot. It was interesting. Our partner program was a really big part of how we grew.
03:13 And then I worked for a PC company for a little while. And this is actually very early on before a lot of this AI stuff it’s coming about now.
03:20 So back in 2019 and I remember even at the time I’ll just mention this as a total aside just because I think it’s relevant for AI.
03:27 I remember I went to Nvidia’s annual user conference and saw a panel on how they were starting to apply GPUs for imaging and reviewing imaging and things like that.
03:38 And at the time it was just called ML primarily. But this was a Linux computer company and a lot of these things are run on Linux and the systems are run on Linux and there had been this like run where people were buying GPUs to mine Bitcoin and stuff.
03:53 And then at some point, they started identifying these other use cases. And I remember we were selling these like server rack units and these really beefed up machines with very heavy GPU compute capability.
04:04 And at the time being like, you know, this is interesting. The most expensive some machines were selling by far have this like they’re just maxing out on GPUs and yeah anyway so that’s come a long way in this six or seven years since then but I really wanted to work for myself by a wife and I were playing
04:21 on starting a family and having kids and so I want to work for myself so I started an agency to do that and was working in the HubSpot space worked as a close partner as well and really really had a great experience working with clothes especially they have an awesome team over there and I really love
04:35 Staley the founder And as I was doing that, we were running CRM ops and then outbound as a service. And that was kind of where outbound sync came from, is that we were using these new outbound tools, trying to help people stay out of the sit band folder and land in the inbox and y’all know these outbound
04:50 tools really talked back to the CRM. And so yeah, we can talk more about the process of kind of getting this out of the ground, but that was how it came about built an MVP, got a little more traction with it, and it just kind of has been building momentum for the last two years.
05:04 But the first version of it I did build with chat GPT. It actually helped me write the API calls and get going So the AI kind of threw out the journey for me.
05:12 It’s a that’s incredible and I haven’t even really talked to you much about outbound sync, but I first had know the challenges that People have with you know any type of outbound campaign and tying it back to the CRM my own incline my own clients included did.
05:32 So I’m really excited to dig into that a little more. What would you say like ultimately was your biggest challenge as you moved from an agency to a SaaS product and then the second question to that kind of to go along with the challenge theme is you know what are you observing as the biggest challenge
05:56 in and the outbound sales process right now. Yeah, well, it’s really been a series of challenges. I think in the very beginning, when you’re running a services business, it’s a very different business model than a software company.
06:14 And so running both at the same time was exceptionally difficult because you have a retainer client who’s paying you thousands of dollars a month and then you have the software user who’s paying you hundreds of dollars a month.
06:29 And so it’s just very difficult to understand how to prioritize your time when you’re just a solo founder and you have these wildly divergent value and time, you know, time to value all these things.
06:41 It was just really, really hard to balance the two. So when I made the decision to go all in on the software product, I think that was a big one.
06:50 It was a big unlock and I joined Tiny Seed, which is a B2B SaaS accelerator for bootstrap SaaS companies. That was a huge unlock.
07:00 So that was one. And then the other one that was kind of during that same time of getting things off the ground was, frankly, just building the product being a non-technical founder.
07:08 It was really hard. And I was using ChatGPT extensively to write API calls and to build it out. But I kept running into technical issues and I just didn’t know how to solve them.
07:19 And so, you know, bring on an engineer and then bring that engineer full time. And then bring on a customer success person, you know, just growing the technical capability of the team.
07:26 I mean, I think that was really big for us, just kind of breaking through. And now to the point where we’re really kind of rocking.
07:31 We have a pretty big part to program. We’ve synced tens of millions of records. I think we’re, I think we’re in a really great spot today.
07:37 But getting here was definitely a lot of work for sure. Yeah. Thanks for sharing that. I, I’ve been working closely with one of my best friends on his SaaS company, and it’s very similar to what you’re saying.
07:51 He struggled a lot with the, are we a SaaS with a service company inside of it? Or are we a true SaaS?
08:01 And we need to get this service piece, agency piece out of here, which he did take that path, but it sounds similar to you, where you kind of was kind of doing both with it.
08:11 He was like, you know, I want to go all in on the SaaS. So that’s really interesting, because it seems to be a common thread about people kind of struggle with that, especially if you came from an agency angle.
08:25 Yeah. So, but I mean, there’s been some benefits. So like, we do some things that when I talk to other software founders, they find unbelievable.
08:34 And maybe these are things that don’t scale, but like we have Slack channels with all of our agency partners. And you know, So we do things that we really try to put service and put the customer first and the user first.
08:44 And I think that’s allowed us to learn and it gives us a little bit of grace. I think, I spend a lot of time focusing on the growth side of the business, not trying to make the product perfect, but trying to understand the market and where there’s pain points in running outbound campaigns and then thinking
08:58 that data back. And so I think the flip side is that if you can just kind of gut it out, there are some really cool things you can do if you understand the product and you understand customers and spend a lot of time talking to them.
09:12 Where I can see a technical founder might get caught, like just writing a lot of code and it’s like, yeah, but nobody’s ever going to use that feature actually.
09:20 But you know, it’s, you know, all you have is a hammer, everything looks like a nail and so you just kind of think like another feature, another feature.
09:25 So yeah, I think there’s pros and cons. Love that. You read the lean startup before? Yeah, I mean, years ago, but yeah, similar concepts, you know, and both sides can get trapped on it.
09:37 So yeah, for sure, totally. Cool, man. I mean, the only thing that’s true for sure is that it’s just hard.
09:43 Yeah, I get that. I love to talk about that while we’re on the topic about outbound sync. So how does outbound sync actually integrate with platforms like HubSpot?
09:58 And do you guys do sales force as well? Yep. Okay. So I’ll have flat and sales force to actually streamline that outbound process.
10:06 What does it look like? Yeah. So I think that the most important thing is that when people are sending these outbound campaigns, there’s just activities, there’s events that are happening.
10:16 Emails are being sent, replies are being generated. Those replies have a disposition, you know, someone’s interested or not. And so what we work on is just syncing those as like an activity feed over into the CRM.
10:28 So we’re logging those, we’re updating or creating contacts and companies in HubSpot or in Salesforce, we’re potentially either using the lead object or we’re using accounts and contacts.
10:38 But the idea is basically just getting that data over. And there’s really three main pain points that we’re solving for.
10:43 It’s like logging and routing those replies, attributing revenue so you can see of outbounds working or not and then maintaining block lists so that you make sure you’re not emailing your assistant customers.
10:53 So when people want their outbound tool to talk to their CRM, those are the three things that they want generally out of our integration.
11:01 So yeah. Awesome. There’s a lot of money in syncing. I know I worked with an organization, actually a company I worked for prior to starting the agency and everything else.
11:18 And it was about HubSpot five years ago. And we had to sink tons of the fusion soft data and striped data.
11:25 And HubSpot payments wasn’t, I don’t even think it. I think it just came out. It didn’t have a good like two-way sink or any connections in it.
11:36 And man, we spent a lot of money. It took about a year, year and a half almost to connect everything and get it synced up.
11:42 So it’s a huge pain point for sure. So, when you get the activities over, could you want to take me through like a typical use case that you’re like, this is by far like the biggest needle mover for our clients when it comes to actually having that data in a home spot?
12:05 Yeah. Well, I think that if you have a sales team And, you know, you’re thinking about the sales rep experience.
12:14 What we find is that people are a lot of times working with agencies to help them build these outbound engines.
12:20 Because these tools are very complicated. So, if you’re talking about using clay and list building and writing copy and creating infrastructure for deliverability, it’s just a lot to manage.
12:32 And so, where users tend to find outbound sync, think is that they want their salespeople to focus on selling. And so they will often lean on an agency, not always, but maybe there will be an internal person to run the campaign infrastructure, but get those emails going out the door.
12:50 And then as replies come in, we’d get that reply, log it, route it to the sales rep. And then potentially do some enrichment, potentially do some AI workflow to like summarize the conversation to date or suggest a reply or something like that to save that rev time so they can just focus on who’s this
13:08 customer, what happened, how can I engage with them? Should I pick up the phone? Should I reach out to them on LinkedIn and connect with them that way or some other network?
13:15 And so that’s where we, I think that’s like the most common thing that we see. The other one is attribution.
13:22 You know, we want to see if app bound is working or not. And so by syncing all this data over, we can see if, for example, contact converts, but they didn’t actually reply to the email, but they ended up converting anyway because they came in through the website, or they forwarded in someone else at 
13:37 the company ends up converting. And we’re going to count that as I don’t find an influence pipeline. But it’s not always this straight line for when that attribution event occurs, right?
13:49 And so building out reporting that’s showing influence pipeline inside of a dashboard inside of hotspot or Salesforce, so that they can see where I’m just having an impact on pipeline generation.
14:01 Those are kind of the two main things and what that looks like kind of step-by-step. Amazing. You mentioned AI in there and it’s kind of summarizing long threads potentially or doing some deep research and figuring out more about the prospects.
14:19 So let’s talk a little bit about AI, what is like your favorite, if you have one tool to pick, you can only keep one AI tool on the planet right now in its current state.
14:29 What would it be? Well, I mean, personally, I just use ChatGPT, I just have it open all the time, it’s basically my co-founder.
14:39 But for our users, where we fit in AI is that we’re actually not an AI company, really. We don’t have an AI product, we don’t actually have AI in our products at all, but what What we do is we operate on the data layer.
14:53 So our job is to get that data into the system of record in exactly the way HubSpot and Salesforce want it so that when people are hooking up the new HubSpot MCP server, or if they’re using Breeze in a workflow, or if they’re using AgentForce, those tools know where to look and everything that we’ve 
15:15 logged is logged correctly and categorized correctly in all the fields and properties are set correctly so that it can do its job.
15:23 So I see us as an enabler. We get that data in there and whether a human is reading it or an LLM is reading it, that’s kind of like a second question.
15:31 I do think that as people start to integrate their systems more in order to apply these use cases, I do think the value of what OutboundSync does goes up because if you don’t have the data in there, you can’t feed into an LLM, which means you can’t have it make decisions for you or automate things or
15:47 save you time. So that’s kind of where our product sits, but for us as a company, I personally use ChatGPT a lot.
15:54 Our dev team uses Curstor and Windsorf. We also use a product called Aikido that uses AI to basically scan vulnerability reports and different things like that that come in from our upstream packages and software that we’re developing and kind of clean that up to save our developers time.
16:11 That’s a really important part of our tech stack from a development perspective. So yeah, those are probably the main things that we’re doing today.
16:19 Awesome. How much time do you think it saves your devs having AI now? I don’t know. Honestly. I mean, we’ve always been lean because we’re small.
16:30 You know, we only work on a few things at a time, and we just kind of get them out the door.
16:33 Yeah. I think that. I think that it’s been really helpful for Salesforce, especially. It’s been Salesforce’s API is very complicated.
16:42 And it’s very difficult to, you know, make sure they’re not creating duplicates and things like that. And so I think that I think they found that to be particularly helpful, just kind of that’s like a little enabling piece along the way as some tricky questions come up.
16:56 I think we’re going to be using it continues to continue to be using it. I think it’s going to be a big part of our part of what we do.
17:01 But I do believe really firmly in, you know, humans being the ones that are reviewing and pushing code into production, I, you know, we make an infrastructure product.
17:09 And so it’s really important that our product works. People expect it to work. And so I think it’s very powerful, but I would say, we’re not like vibe coding things into production.
17:20 That’s good. Yeah. Yeah. The thing I always speak a lot about with when it comes to AI is context. And the context is everything.
17:29 Regardless if it’s development, if it’s copyrighting, there still has to be someone to put the guardrails on to provide the context.
17:37 And I don’t think that’s going away any time soon, especially when it comes to systems, because I think there’s still, at least from what I’ve seen, there’s still this missing layer of UX and design and the actual structure of the code and the system itself, when it comes those vibe coding stuff.
18:05 So you can get it to look pretty good, you can get it to function pretty good. But a lot of times when a real developer, you know, a real talented developer would go in there, they would wonder what in the world was going on and who wrote this code.
18:22 So yeah, it’s interesting. Yeah, I think it’s very exciting. And I think that I think there’s also risks associated with it.
18:31 I mean, we have, I mean, we have SOC 2 type II attestations. So we went through a pretty rigorous process to like have the way that we develop and ship code, you know, we follow a process.
18:44 We follow rules around that because because we write the system of record and we have access. We have read right access to systems of record for companies that are pretty big.
18:54 You know, So they have thousands of employees mid-market companies that have made promises to their customers and to their partners about how they manage their environment.
19:05 And so, you know, I think we take our responsibility pretty seriously there and that’s why we invested a lot in security.
19:11 Even though we’re a small team, you know, that was like one of the early things that we did when we joined TinySeed as it’s invested in that and I think it’s been an absolutely worthwhile and that’s something that we’re always going to, I would say probably over-invest in relative to our size because
19:24 of the nature of what we do. It’s just a really important thing. Yeah. What would you say when it comes to AI, is there anything you’re like, not fear, but I?
19:35 What scares you a little bit about AI? If anything. Well, I think that people are overconfident in the outputs that they’re getting from it sometimes.
19:46 I mean, I think it’s exceptional, but you know, I’ve personally had very bad experiences with customer support from AI tools where it’s telling me things that are not true to do.
20:02 And I mean, I’m a power user of software. I’m very familiar with the tools that we use. And so if I’m in, if I’m in a support environment, it’s what I found is that every single time I’ve had one of these like really frustrating experiences, it’s because I’ve found a bug.
20:17 And so the AI doesn’t know it’s a bug. And these are companies that don’t have status pages, They don’t have, they don’t have a report, you know, so, I mean, I guess the answer I would say as well, it just needs more and more and more and more context and feed it more and feed it more and feed it more
20:29 and there’s this infinite, but I don’t know. Anyway, so that is a thing that I’ve had a negative experience on as a user, as a customer, where it’s like telling me to do something and I know it’s not going to work, but had I listened, I would have wasted like 15 minutes troubleshooting something that
20:44 had nothing to do with the problem. So that’s kind of frustrating to me. I think that like having a human in the loop with AI support is a totally different thing.
20:51 Having a developer in the loop with AI support is a totally different thing. So yeah, I don’t know. I mean, I think AI is great.
21:00 I think it’s not going anywhere, but I think that people are overconfident in some of these uses. And I think they’re not thinking about the consequences.
21:06 And so having an expert on your team or an external resource that you bring in to help you build out agents or workflow, agentic workflows or whatever, just to make sure that you have those guardrails.
21:16 And for someone that knows enough to say, we’re asking too much here, just have it do this part of the job and that’ll save you enough time.
21:24 But you don’t have to automate the whole thing. I think that teams and companies need to do that. Otherwise, they just might really frustrate their customers and I think that’s not a good thing.
21:36 Yeah, I find it amplifies laziness a lot of people, which is scary sometimes because Because if someone is lazy and now they have a tool that can help them do their work and they already were lazy, the odds of them checking the work is unlikely.
21:57 And I think that’s where a lot of those issues arise is it’s information that they should have checked before sending it, maybe they got it from Chat GPT and they didn’t check it.
22:07 And that one scares me quite a bit, you know, because you do have parameters around, you know, how people use that.
22:17 A question for you, because we have been, well, I won’t tell you yet, I wanna hear your answer. Have you seen a improvement or degradation of chat GBT’s responses and use or the last, let’s say, four weeks, six weeks.
22:41 Yeah, it depends. I’ve definitely, I’ve overall, I would say it’s been relatively the same, but I have had these like outlier response things that have been coming back, where I’m like, this is annoying.
22:57 This doesn’t work. This isn’t useful. That have been a little bit frustrating. And that’s kind of like part of why I say like, you got to have somebody check in this stuff, because it’s like, you know, so yeah, I would say it’s a little spotty.
23:10 I’ve been trying different models. People have been talking a lot about clawed, so I’ve been looking at, I mean, I tried clawed like years ago, but I didn’t ever really like fully adopted it for whatever reason.
23:20 I just got it more into chat GPT. So I think having multiple things that you can lean on is probably an interesting thing, but I don’t have enough.
23:27 I haven’t been doing that yet. I want to because of MCP. I want to be able to try out MCP servers and, you know, anthropic develop that.
23:33 So, but yeah, I’ve noticed recently like some little things that are off, but I don’t know if it if in these chat windows If the context window is like if it’s just too much going on I need to kind of start fresh or what I try to manage the memory But that feels like very manual, but I go into the memory
23:47 sometimes and I delete things that I think aren’t helpful or outdated But I don’t know. I mean this stuff’s happening so fast.
23:53 I don’t even necessarily consider myself Like an overall AI expert, but I do think I know how to use it to start and grow a company but I wouldn’t enjoy myself an AI expert in and of itself if that makes sense.
24:04 I mean, that’s what this podcast really is about, right? It’s like getting the insights and opinions from different people using AI at different levels.
24:12 You know, we’ve had guests that are launching AI companies. We’ve had guests that had never used AI before. You know, and then I’m somewhere in the middle where we’re doing some really cool AI agents, you know, and my pretty good, you know, chat GPT user.
24:27 But I definitely am not the developer, you know, of, you know, any type, anything ML, but we have, at least I have been seeing a huge uptick in like hallucinations in chat GPT.
24:40 I’m really weird ones where like, I’m like, please don’t do this. And then it like does that exact thing. And I’m like, hey, like, I just asked you not to do this, you know, and it’s just get stuck in this loop.
24:51 And I was pretty recent. Yeah, so I’ve been asking people lately and I put an Instagram post up and said, hey, if you guys, you know, seen an uptick in this and it was the most popular Instagram story I’ve posted in like the last six months.
25:06 I had like, I want to say like 35 responses to it like, yes, we’re doing the same thing. It’s crazy, so it’s really interesting to me that AI could decide almost at any time to to go run off into its own little world.
25:24 Like, you know, it could, in theory, decide to just degrade it. So, you know, like, it’s a weird concept or to like manipulate, if it wanted to, and people wouldn’t even know.
25:41 There was actually a post, I think Joe Rogan actually reposted it, but it was one of our buddies that did an I come back or he did a thing on the Reddit.
25:51 There was a company that did like a Reddit study and like, you know, showed that it was six times more effective at convincing people To change their opinion on something So yeah, it’s pretty wild As a non-technical founder without a you know, I’m a co-founder part like a technical co-founder Chat TBC
26:11 is basically my co-founder. Amazing. So with that I guess Yes, what were some technical hurdles that you ran into trying to build a software, especially before you brought in engineers and there’s initial stages, what were some of the biggest technical challenges, how would you get around those, how 
26:33 would you solve that? Yeah, I mean, this data sync problem is just a lot harder, I think, than people realize on the surface.
26:40 I think people think, oh, okay, well, I could use make, I could use Zapier, and I thought that too, the first version of OutboundSync, I built with make.com and I was like, oh yeah, this is going to be so easy.
26:51 And then I was like literally three in the morning and I’m troubleshooting why this API call isn’t working. So I mean, it’s just like getting data to format and you’re going between like, you know, you’ve got email data, which is HTML and then you’re switching that to something that has to be JSON compatible
27:05 for an API call. And I can definitely help, it can get you a certain amount of the way, but there were certain problems that just a regular or not regular, I think our engineer effort is exceptional for our size, but you know, someone who knows software is like, oh yeah, no problem, fix, done, where 
27:23 it’s like holy cow. I spent like two weeks losing sleep over that, trying to figure that out, but it was worth it because you know, I basically was like, do customers have this problem?
27:36 I will just kind of get the basic as basic version of I can out the door. But there was just these weird nooks and crannies around data compatibility and data formatting and stuff like that that were just really difficult.
27:46 And then like, you know, in a perfect world, the problem I think I have a lot with a lot of this AI dependent stuff is that there’s very limited error handling.
27:58 So if you build, if you use an AI to like read some API documentation and then you write something based on that, you know, there’s this assumption that the documentation that you’re it is correct, but what if it’s not, or what if the data that you’re getting is inconsistent, you know, and then like,
28:15 I mean, the deal with the thing with these support use cases that I’m in is that the company has a bug, the documentation doesn’t reflect that.
28:22 I don’t know if they didn’t test it or they did test it or what. The status page does not reflect that there’s a bug.
28:27 Their engineering team is saying there’s not a bug, but there is a bug and I can see it and I’m proving it and I’m showing it.
28:32 And then they eventually were like, oh yeah, okay, we’ve been able to replicate this. So, like, there’s a conflict of interest problem for an LLM in this situation, like, who does it believe?
28:44 Yeah. Does it believe the company who’s paying it or does it believe the user? And historically, in some of these use cases, the customer success, they actually have like a little bit of an antagonistic relationship sometimes, sometimes with internal teams.
28:56 Their job is to be the voice of the customer, and sometimes their job is to be a little annoying to internal teams and to say, hey, you know what, this is a problem.
29:02 I know you say it’s fine, but customers are trying us, it’s not. And when you automate all that, like, what are you losing?
29:09 And so, you know, that’s a, that’s a challenge. It’s led to challenges. I mean, when, when, when I’m dealing with other companies, when it’s internal, we are able to flatten that out a little bit more, but yeah, like assuming that the data will always be what they say it will be if you’re building an
29:24 integration product, like that’s just not true. Like we just dealt with a bug where every once in a while randomly, a timestamp, instead of having three millisecond digits on it, it has two.
29:33 No idea why, but Salesforce requires three. And so that would break these API calls from being successful in Salesforce. And it’s like why did the upstream data provider do that?
29:43 I have no idea, it’s not documented anywhere. So I think when things go wrong, they can go catastrophically wrong without some person checking and monitoring and implementing fixes.
29:55 And if it’s dependent on the information that’s provided to it, that like you said earlier, like it amplifies laziness, is it’s just working off of what it has.
30:04 And so if the company has imperfect or incomplete documentation, that’s amplified by whatever LLM is reading that documentation, you know?
30:12 So. And then regurgitation, because it’s relearning as other people are posting about it, you know? Yeah. Yeah, it can become a problem pretty quickly.
30:22 And that’s what agentic is really interesting to me because, you know, the question would be like, hey, is there a world where, you know, you know, a genetic, for example, might solve that problem because you can say, hey, go do this thing, right?
30:35 And it’ll go do it and see the error. And they’re like, oh, yeah, this is happening. And it’s not a documentation.
30:41 So, you know, in the near future, there might be, you know, it might be a little better when it comes to things like that, but I have no idea.
30:48 And it happens. Well, I mean, I’m optimistic overall, I’m very optimistic about the use cases and that things will continue to get better.
30:56 You know, but I think to say everything is great now I was just not my experience. For sure. Yeah. And to your point with the bugs, I mean, it’s so true because there’s two specific cases with our clients where we were using this one dialer that were kind of like a partner of when we use a lot and I 
31:14 was feeding HubSpot numbers into it. And the numbers were supposed to be like international format, no spaces, whatever. I find out that HubSpot actually wasn’t putting them in that format.
31:26 And so the data going in was wrong. And then I find out that it was causing us not to be able to delete the numbers out of like removal from the diver.
31:36 Now, but then I was like, wait a second. If I put it in wrong and I pull it out wrong, like technically that’s what went in.
31:44 So I go, I mean, I so much trying to figure this out. I find out basically what’s happening is they had, you know, a piece of Python code, the diver did.
31:52 So when you put the wrong phone number in, and it’s actually fixing the phone number. Phone. When you delete the phone number, they don’t have that IP code on it.
32:03 So they would remove all the spaces and clean up the phone number when you put it in, and then I would go to remove it with the same number that I put in, but it’s been cleaned.
32:11 So I can’t pull it back out. So it is those small things where, I mean, had I not just sat there and looked at so many examples I was like, why is he actually, how was this possible?
32:24 It shouldn’t be possible. AI would have never figured that out, at least to current state. There’s no way. So yeah, and then another one was pretty simple.
32:36 I had to do with HubSpot 2, which was, they were pulling an IP country. And the IP country was supposed to be correct based on all this, they did a lot of work to get it right.
32:49 And I was supposed to use it for another workflow, had I assumed that it was correct, my whole work would have been broken, but I went in there and I’m like, sorry, look, are these actually right?
33:01 Turns out like 98% of them were wrong. So that human touch is super, super important, but cool. Well, I know we’re coming up on time, man.
33:13 I want to be respectful of your time. I really appreciate you, you know, setting this side. One question I want to ask, and for my listeners to know, what do you see as like the future of outbound sync?
33:26 And then what is the feature that you’re most excited about in the product roadmap for you guys over the next few months, years, whatever, and maybe?
33:37 Yeah, well, I tend to live life a few weeks at a time around here. We don’t have long-term roadmaps, because things have changed quickly.
33:47 But, you know, where big picture where I think things are going is the same problem that we’ve identified with email as a channel is also true for other channels, like, like, like, DIN or, you know, B2B social and phone.
33:59 And so, you know, we found the integrations in these other channels are not great. And then, beyond that, there’s really a lack of ability to have, like, essentially a switchboard between these channels.
34:10 So, if you think about what OutboundSync is doing, we essentially attach or connect these point solutions to main systems of record, where there’s workflow automation, there’s assignment of things, there’s like a system of action, there is potentially AI being fed off of this pool of data.
34:28 We sit there in the middle between a point solution and the CRM, but what we’re finding is that there’s opportunity potentially to help manage workflow between point solutions.
34:37 So, if I make a call and someone answers, I want to automatically stop my email campaign to that person, right?
34:46 Because I don’t want to email them because they just answered, I just talked to them, like we’re good. They’re now moving forward in a pipeline, you know, they’re going to create a new dealer, convert the lead to an SQL or whatever.
34:57 And so, that’s where I think there’s some really interesting opportunity for us to help teams out and do some interesting things.
35:03 So, that’s what we’re thinking about right now, and I think there’s a lot of opportunity there. And then, you know, we’re looking at having our own API for our own data that’s sitting in the middle.
35:13 We’re sitting on over 25 million records now. And so, and that’s growing every day. And so it’s like if we make that data available for tools kind of between tools, that also gives us the ability to potentially have our own MCP server and to make that data really accessible for people.
35:29 So I mean, I think while we don’t have specific plans, I really think over the next three to six months, that’s going to be what we’re really thinking about.
35:36 Amazing. That’s a pretty incredible that the other tools I’m excited to see what you guys come up with. You know, we have, I have probably like five that come top of mind right away that there’s just very poor communication, you know, I mean, like I’m barely able to create a, you know, a call record 
35:52 for a kid to go in and it’s still a little sketchy, you know, um, you know, HubSpot’s really, really buggy when it comes to like, you know, the information It needs and so I’m excited to see that and I might be able to connect you with some some great people too that you know, you could have some great
36:10 success. You know with some of those future products that you’re integrating. But awesome man. Well for our listeners, what’s the best way to learn more about out outbound sync and get in touch with you specifically?
36:24 Is it linked in YouTube? Yeah, for sure. So just outboundsync.com, s-y-n-c C. And I’m on LinkedIn. My name is Harris Kenny.
36:35 Feel free to look me up there. We do have a YouTube channel. If you like YouTube, you can look up outboundsync there.
36:39 We have a YouTube channel as well. And yeah, man. And yeah, appreciate the chat. You know, you’re an operator. You’re doing this stuff every day.
36:46 And so it’s fun to hear about it from your perspective and thanks for having me on the pod. Yeah. Of course, man, we’ll put your links in the description of the podcast for everyone and go check it out.
36:59 I know I definitely have some people that struggle with these same things. So it’s cool to find solutions and hear your perspective on AI and you know how you’re using it to help build your business.
37:10 So it’s pretty amazing. And yeah, thanks for being on the show and we’ll see you next time. Okay, man. Talk soon.
37:18 Thanks. Yes.