Can AI Improve Trust Between Political Campaigns And Voters?

[00:00:00] And if you are running a business right now, you may have noticed there's a quiet shift happening. One that most people are still underestimating. And that is your company doesn't live inside your network anymore. It lives inside the browser. That's where your SaaS apps sit. That's where your data moves. And increasingly, that's where attackers are focusing their attention.

[00:00:23] So NordLayer has just launched its new business browser. And it's designed specifically for small and medium-sized companies that need visibility and control without the overhead of enterprise security tools. What I like here is the balance. You get advanced protection, better compliance, and full visibility into how your team is working online, but without slowing anyone down or forcing them to learn anything new.

[00:00:50] So it feels like a practical step forward rather than another security layer that adds friction. So if you want to see more about how it works, please head over to nordlayer.com slash browser and check it out and let me know your thoughts. But now, on with today's show.

[00:01:11] What if the biggest breakthrough in political texting has nothing to do with sending more messages and everything to do with finally listening to when people reply? Well, we've got a different one today. I'm going to be joined by Tom Carroll and Andrew Missy. They're the co-founders of Convos. And I love this conversation because it focuses on a part of political communication that most people rarely think about. Campaigns.

[00:01:38] Because they've gotten very good at blasting messages out at scale. But the harder part becomes the moment a voter texts back. Texts with a question, a concern, or even a bit of confusion. That is where things get messy, expensive, and very human very quickly. And my guests are on a mission to fix this. They have built convos to help campaigns move beyond one-way broadcast messaging and into something far more conversational.

[00:02:05] So we will talk about how they are helping teams manage thousands of replies. How sentiment and urgency scoring can help surface the messages that matter most. And why a short question can often do far more than a long block of campaign copy ever could. But what stands out to me is this isn't a story about automating people out of processes. It's about using technology to make room for better engagement, faster responses, and more meaningful interactions at scale.

[00:02:35] And with Tom and Andrew having been friends since kindergarten, there is also a brilliant founder story running through this episode that makes the whole conversation feel so much more enjoyable. So if you are interested in where AI, communication, and real-world engagement collide, this is a fascinating one. So let me introduce you to them now. So a massive warm welcome to the show, Tom. Can you tell everyone listening a little about who you are and what you do?

[00:03:05] Yeah, absolutely. My name is Tom Carroll, and I am the co-founder of Convos. So really my background is in finance and entrepreneurship. Did a little bit in e-commerce, a little bit of real estate. But then sat down one day and started working with Andrew on building what originally was called PubSent and now is Convos.

[00:03:25] And really the goal here is that we're adding personalized conversations to political texting and really turning that one-way broadcast into a conversation that the campaign can have at scale using an AI agent. And that's the platform that really, I mean, Andrew built, which is Convos. That's incredibly cool. And you mentioned Andrew. Now I was going to surprise everyone, but the cat is out of the bag.

[00:03:53] Andrew, can you tell everyone a little about who you are and what you do too? Absolutely. I'm Andrew Masset. I'm the CTO and co-founder of Convos. And yeah, definitely the technical side. I've been a software engineer for a little over a decade and me and Tom, fun fact, have actually been best friends since kindergarten. And so we've known each other a very long time and we've done a couple of different things together, but this has been one of the most exciting and really excited to talk about it and tell you where we're going.

[00:04:22] Oh, what an incredibly cool backstory. I absolutely love that. What I wanted to talk with you about today is political SMS because it has been around for years, but I know you argue that the real problem was never about just sending messages. It was actually handling what happens after voters reply. So what was fundamentally broken in that second half of the process that you tried to fix that?

[00:04:46] Yeah, so every campaigner knows that if you have a small campaign and you're able to send replies and talk to people, you really get a lot more out of that. You really learn a lot more about what the constituents and what the voters are looking for. But the problem is, is that any point of scale, really any, definitely any statewide level, you have it where you're sending out hundreds of thousands of texts.

[00:05:12] You can get tens of thousands of people replying and it's just, it's impossible to have the resources to build out, to respond to each and every one of those. And even if you did get a massive set of volunteers to do responses, to then gather all that data and really understand what the people are talking about and quickly, that's just where that broke down previously and where we've been able to build in.

[00:05:36] And so what we're able to do is respond, you know, in every, to every single message in about 12 to 15 seconds, and then gather all that like sentiment and the topic data and really just everything that people are talking about and display that live to the campaign. And of course it is such a busy time of year around the political scene at the moment. And one of the things that stood out to me about Convotes is how you're focusing on conversation rather than sheer volume and scale.

[00:06:06] So when you take that approach, how does that change the way that campaigns think about engagement? And ultimately what does a successful interaction actually look like? It really changes that first message. And I think that's kind of the, there's art and there's science to this. And really a bit of the art is trying to lead these campaigns into saying, hey, how about that first message instead of a large block of text and a link?

[00:06:32] While that certainly has a place in the spectrum of political messaging, let's make that into a short question. Let's just ask the people their opinion on a hot topic issue or what issue is top of mind. And then from there, the person responds. And how it works, and I'm sure Andrew will dive in a little bit more in detail, is that there's a closed knowledge base that the AI has access to.

[00:06:56] But from that, it can just really answer the question exactly how the campaign would like based on the information provided by the campaign. And so you're getting, you know, questions about, let's say, a nonprofit's history, a political candidate's voting history if they've already been in office. Everything like that, the AI can then share in a, just in a timely manner, exactly tailored to what the person's interested in.

[00:07:22] And that's that point of this becomes, you know, not only a tool for the campaign, but a tool for the people that they can then use to really educate themselves. And one of the things that stands out is you're also seeing reply rates significantly higher than some of the traditional benchmarks out there. So alongside, especially alongside relatively low opt-out rates. So when you put all that into the mix, what do you think is driving this shift?

[00:07:48] And what are campaigns doing differently when they treat messaging as more of a dialogue than a monologue? Yeah, I think it's really just, I keep going back to it, it's that listening. It's that once you use our platform once and then you start getting data being like, okay, this is what people are interested in. This is what they're talking about. Let's talk about that. And it's almost simple in what you think about, but it really is a complex thing to reach as far as the technology goes.

[00:08:18] I mean, it's expensive polling. It's really just extensive canvassing. It's gathering all of your door-to-door data and compiling it down. But when you can do that via a text message and just send out a large message saying, hey, what do you guys think about this? People respond, bring up the issues, and then you tailor your messaging to that with the next send. That's what's really going to increase engagement because it just makes sense at a human level.

[00:08:45] People are engaging what they want to engage in and what they're interested in. Another one of the most interesting aspects of your platform here is sentiment and urgency scoring. So how do you ensure those signals are accurate enough to guide real campaign decisions, especially when the stakes can be so high? Yeah, I'll jump in on this one. So the way we actually do our sentiment scoring, and fun fact, we're actually in the process of changing it from sentiment more to alignment.

[00:09:15] Because what we found is what campaigns want to know is when they're reaching out to these people, how aligned are the voters with their campaign? And that's really the score that they're looking for. And the way that we're able to do that is when we're analyzing chats that have occurred, we're actually taking the information that the campaigns have given us about their candidate and cross-checking that with the conversation. So again, Tom said something about art and science. This is also a bit of that.

[00:09:42] But it is kind of seeing how the voter relates to that information and then giving them a relative score based on that. We also do this with topics that they talked about. So if somebody has a specific issue or a specific interest on immigration and their interest aligns with the campaigns, we'll actually surface that to the campaign as well and say, hey, they were aligned with your views on immigration. And the campaigns really find that information very valuable.

[00:10:12] And you've also shared examples of where this approach correlates with increased turnout and increased vote share. So tell me more about that. I mean, how confident can campaigns be in attributing some of those outcomes to conversational SMS versus broader political or social factors? And one of the reasons I say that is I think just every tech project now or anything involving technology has got that big ROI question mark above it. So tell me more about some of that ROI that you're seeing here.

[00:10:42] Yeah, it's a little bit of column A, column B as far as there's some data there and there's some anecdotal evidence. And so, I mean, with any campaign, they have to use every tool at their disposal to raise the vote share. And so every tool is going to definitely contribute. So, you know, it's definitely not something where you can say, you know, the vote share went up by 10 percent. That was all us.

[00:11:07] But I do like to say is that we have a lot of evidence of, you know, you have a conversation and they can see every chat that happens where somebody says, oh, I didn't know that event was happening. And then you show up to the event and that person is there. And they said, oh, hey, I got your text. Really excited to see what happens here. Really excited to ask the candidates some questions at this open town. There's different words for it.

[00:11:33] But it's there's a lot of different ways that we've seen people, you know, start the process of getting interested. And I think that leads them down the rabbit hole of them learning about the candidate, getting information from multiple different ways and more just getting involved in the political process where that might have felt a little bit more overwhelming. A lot of work we do is in get out the vote. And so in the past, it might have been, hey, make sure you vote today and then leave it there.

[00:12:02] We offer a tool that says, hey, the vote is coming up tomorrow. Do you have any questions? And somebody says, hey, where is my voting place? Where should I go? We could say, OK, this is a link provided by the campaign that can tell you where your voting location is. Do I need to bring anything? What are the laws? And the laws change in every state. And so the campaigns will oftentimes load in the voter laws for their state and say, OK, you need to bring this and this or no, you can do this.

[00:12:32] I mean, we just had a conversation where somebody said, what should I wear to the polls? And so instead of a volunteer taking that question, we were able to say, doesn't matter. Just come out and make sure you vote. And so it's, you know, it's just kind of it's a little harder to facilitate exactly what the ROI is. But we're definitely seeing awesome results as far as people that maybe weren't previously looking to get involved are learning a lot more and taking that step.

[00:13:03] Big thank you to Denodo for supporting the Tech Talks network and making these conversations possible. Because when your lake house stores the data, the real challenge is getting that data where it needs to go and faster. And your lake house stores the data, but Denodo helps deliver it faster.

[00:13:24] So with real time access, built in governance and a business ready data marketplace, Denodo can help your teams unlock insights without costly duplication. And you can learn more by simply visiting denodo.com. I think there's always a very fine line when introducing automation or any technology into political communication. Again, especially with those stakes being so high.

[00:13:49] So how do you get that balance right between efficiency and scale, but keeping authenticity and avoid creating interactions that feel artificial or manipulative? And if we've anyone that scrolled down their LinkedIn news feed, for example, recently, it feels like every single post is written by the same person. There's a lack of authenticity there because of AI. So how do you get that balance right? Yeah, I think it's definitely a balance. And it's a great question.

[00:14:14] I think being authentic is less to do about sounding human and more about being honest and transparent about, you know, your views and your policies. And that's something that our AI does very well, I would say. Because of the guardrails that we have in place, the source points that these candidates give are the only things that the AI can use when it's having its conversations.

[00:14:38] So these voters are getting directly from the source, the information that they're looking for with no fluff. And, you know, no matter how, what time of day they message it, they're always going to get the same quality of response, which might not be the case in for human volunteers. I would also say, additionally, we've seen a lot of success with campaigns that write a starter message where they actually disclose in the beginning that it's a virtual assistant.

[00:15:05] And people actually sometimes have longer, more thoughtful conversations. Maybe it's because they don't feel like they're being judged by a human and they just feel like this is a great opportunity to get information. But it's been a very interesting thing seeing that balance with disclosing even that it's an agent.

[00:15:25] I just wanted to hop in there and really just double down on what Andrew just said, because I think it's so important, is that the transparency that we look for really helps every interaction. Where it's we're not trying to trick people. We encourage campaigns to disclose with the first message. If anybody asks during the conversation, it always responds saying, yes, this is an AI agent. I'm here to help you. This is a tool for you. And so I think it's, you know, really what are you saying there?

[00:15:56] It's that once you reach past that level of transparency where everybody knows what's going on, you're not looking for the exact personality and tone, which although I do think we do a great job doing, you're just more so looking for information.

[00:16:11] And as we look ahead and conversational platforms become more sophisticated, are there any safeguards that you think need to be in place to ensure this kind of technology strengthens democratic engagement rather than undermining trust? And we've seen a lot of deep fakes and things out there. There's a suspicion around all things technology sometimes. Any safeguards with it being political that you think need to be in place here?

[00:16:36] It depends on every part of the process, and you really have to have those safeguards in every part. So it's you start off with step one. Anytime someone sends a message that indicates at all that they don't want to talk, you know, it traditionally you have to send S-T-O-P and that's it. And if you misspell it, then you're still on the list.

[00:16:57] For us, you can send S-T-I-P, you can say, I don't want text anymore, don't reach out, whatever plethora of ways to people phrase it, and you're immediately treated the same as a stop message. So, I mean, step one is just respecting people's decisions and whether they want to be contacted or not. Step two is, you know, with this type of mass messaging, detecting if there's ever a threat to the campaign, if there's any troubling messaging, and then letting the campaign know that automatically and immediately.

[00:17:27] And then really past that, the guardrails just being making sure we're only following the campaign's voice, that they're using information that the candidates have put out publicly. You know, this is previous statements. This is previous speeches. This is their public website. It's that information that is publicly accessible and verified by the campaign really just condensed into one knowledge base.

[00:17:53] And if you can, instead of, you know, asking maybe a generalized LLM that's scanning the entire internet, but instead just have access to that pure information that you actually need, I think it'll build up a lot of trust and build up the guardrails that people need to learn at the speed that they want to, of really find out exactly the information on the policies they care about.

[00:18:47] Yeah. Super great, because you always want consistent messaging whenever you're in a high-stakes environment like politics. And Andrew, looking towards the future there, anything that particularly excites you about new technologies, how the space is going to evolve, et cetera? Absolutely. I mean, we are built on the latest and greatest of AI technology.

[00:19:10] So whenever, you know, larger companies come out with better products, we're always looking to implement those and run them through their paces. And I think as the context evolves around that space and these LLMs are able to follow instructions better and better, we're really going to be able to squeeze a lot more personalization and persuasion ability out of them as well.

[00:19:34] And just to touch on that, I mean, it's a lot of fun, the conversations me and Andrew have putting every single model through. I mean, it's, I feel like everybody oftentimes uses these models for mass data and really it's calculation, maybe it's coding abilities. And, you know, that's talked about very often. I think it's been a, even a selling point, a lot of these major models. The interesting thing is that we are really about, you know, a large context window, but also the personalization, like Andrew said.

[00:20:03] So every time the model comes out, being able to put it through hundreds of thousands of messages and seeing exactly our response is very interesting to see. And, you know, it's only gotten better and better. Well, I love what you're doing here, especially having intelligent conversations at SCALA. We've been able to talk politics without talking blue or red as well, which is incredibly refreshing. We're all still friends at the end of it and it's polarized world that we find ourselves.

[00:20:29] So for anybody listening wanting to dig a little bit deeper on anything that we're talking about here, wanting to find out more information, connect with you or your team or find out more. Where would you like me to send everyone? They can stop right by our site at convos.ai, C-O-N-V-O-S dot A-I. On our site, you know, we have a lot of features, a lot of descriptions on everything I've been talking about.

[00:20:52] And also you can book a demo with us, sit down with me and Andrew, set time aside to really see how we can work with your campaign. Oftentimes people from the demo will set up a test account at no charge. So you can go in, set up your own source points, your knowledge base and, you know, have your team start texting an agent and really learn what those conversations look like. And at the end of every episode, I always say technology works best when it brings people together.

[00:21:18] And also that this podcast is indeed a dialogue, not a monologue. And I'm looking at some of the things you've been putting out there. You'll be exactly the same here. The epitome of this communication should be a conversation, not a broadcast. And I think now more than ever, using technology as an opportunity to understand, engage and build meaningful connections. This is exactly where we need to be heading. So kudos for everything that you've created here.

[00:21:43] And more than anything, thank you for sitting down and sharing that story that began when you were both in kindergarten. And look at the great things you're doing now. Thanks for joining me today. Thank you so much, Neil. Thank you, Neil. This is great. What I enjoyed most about this conversation is that Tom and Andrew are not trying to solve the loudest part of political messaging. They're solving the part that usually gets ignored. Anyone can talk about scale. Anyone can talk about sending volume.

[00:22:08] But what really matters is when someone replies with a genuine question, a practical concern or a moment of hesitation. That is where trust is either built or lost. And what Convo seems to understand is that the reply is not the problem. The reply is the opportunity. So I think there's a wider lesson here that goes well beyond politics. Whether you are building for campaigns, customers or communities, as people, we want to feel heard. We want clarity. We want relevance.

[00:22:38] And we want answers that actually meet us wherever we are. And technology absolutely can help with that. But only when it's being used to support better human connection rather than just replace it with noise. And I like the honesty in this conversation around transparency. They're not pretending technology is something that it's not. They are leaning into clear guardrails, clear social material and clear disclosure. And in a world where trust feels increasingly fragile, that matters a lot.

[00:23:07] But after listening today, I'd love to know what you think. As conversational AI becomes a bigger part of how organizations communicate, how will it help create better engagement? Or will the real difference still come down to how we actually listen? As always, techtalksnetwork.com. Lots of ways that you can message me on there and browse through 4,000 interviews. But that's it for today. So thank you for listening as always. And I'll speak to you all again very soon. Bye for now. Bye.