Pod Insights MarTech Podcast ™ // Marketing + Technology = Business Growth AI-Led Social: Going From Listening To Interaction About this episode In this episode of the MarTech Podcast, host Benjamin Shapiro sits down with Matthew McGrory, Co-founder and CEO of Arwen AI, to explore how artificial intelligence is transforming social media management. They discuss the growing importance of real-time brand engagement, the shift from broadcasting to conversational marketing, and how AI tools like retrieval-augmented generation are enabling personalized customer interactions at scale. Matthew shares insights into identifying meaningful social conversations, managing toxicity, and amplifying positive sentiment to drive brand engagement. Actionable strategies include leveraging sentiment analysis, implementing clear community guidelines, and combining AI with human oversight to maintain authenticity and protect brand reputation. Show NotesConnect With: Matthew McGrory: Website // LinkedInThe MarTech Podcast: Email // LinkedIn // TwitterBenjamin Shapiro: Website // LinkedIn // TwitterSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
Transcript 0:00 Speaker 00 The Martech Podcast is a proud member of the I Hear Everything Podcast Network. Looking to launch or scale your podcast? I Hear Everything delivers podcast production, growth and monetization solutions that transform your words into profit. Ready to give your brand a voice? Then visit iHereEverything.com.
0:25 Speaker 01 From advertising to software as a service to data, across all of our programs and clients, we've seen a 55% to 65% open rate.
0:36 Speaker 02 Getting brands authentically integrated into content performs better than TV advertising.
0:42 Speaker 03 Typical life span of an article is about 24 to 36 hours.
0:46 Speaker 01 If we're reaching out to the right person with the right message and a clear call to action, then it's just a matter of timing.
0:53 Speaker 00 Welcome to the Martech Podcast, a member of the I Hear Everything Podcast Network. In this podcast, you'll hear the stories of world-class marketers that use technology to drive business results and achieve career success. Here's the host of the Martech Podcast, Benjamin Shapiro.
1:15 Speaker 04 73% of users say that if a brand doesn't respond immediately on social media, they'll buy from a competitor, according to Sprout Social's 2025 Social Content Benchmark Report. Turns out that your time to response might actually be more important than what you say to the people who engage with your brand on social media. Most brands use their social channels to advertise. Some of them use them to listen. Very few use them to interact. So how can you transform your social channels from a passive monitoring channel into a hub for meaningful customer conversations? I'm Benjamin Shapiro and joining me today is Matthew McGrory, the co-founder and CEO
1:55 Speaker 04 of arwen.ai, which helps marketing teams amplify positive social interactions. And today, Matthew is going to explain how AI is revolutionizing social media management. Matthew, welcome to the Martech Podcast.
2:10 Speaker 05 Hi Benjamin, thanks for having me.
2:12 Speaker 04 Very excited to have you here, excited to dig into some of the changes that artificial intelligence is having in social media, specifically moving it from being a passive monitoring tool to a place where proactive conversations can happen. Tell me a little bit about how AI is transforming social media to move it from passive monitoring to more conversational.
2:35 Speaker 05 Well, yeah, it's speed of light stuff and it's changing month to month, but what we are really seeing is brands are finally realizing that social media is about being social, it's about having conversations, it's a great place to interact with your customers and your prospects. And what AI is finally allowing people to do is find the buying signals, find the golden nuggets of conversations that they want to engage with and find the people that are positive for their brand and allowing them to kind of automate some of those traditionally very labor intensive processes and get on top of those overwhelming expectations we're seeing from a lot of social media users that they want, they want to talk to people right
3:17 Speaker 05 now and they want instant reaction to what's being said and what they're saying and what they're asking.
3:22 Speaker 04 It's funny because I think of social media in the layman's term of the first challenge is figuring out what you're going to post, right? That seems to be the biggest challenge for most social media marketers is what am I going to put out into the world that's going to capture attention and hopefully create some engagement? And it seems like you're saying that the engagement actually is really the priority. Talk me through that.
3:46 Speaker 05 Yeah, so it's a two way street. Old school advertising on TV was about sending my message, broadcasting the message out to the world. That's kind of 1990s technology. We're in the 2020s now and we've got the opportunity to say what we want to say, people respond, we engage in a conversation and that makes people feel close to the brand, close to the product, close to the channel, close to the subscription that we want to put them on to, and that feels like publishing. Is half the battle and where people are really waking up mature social media teams is they get that they have to interact. You talked about the Sprout Social report in one of the other reports we looked at.
4:29 Speaker 05 They reviewed, it was a 2024 report that only 7% of brands, and we're talking the top 500 B2C brands, US and Europe. Only 7% are actually engaging meaningfully in responding to people when they talk to them on social media.
4:43 Speaker 04 So it seems like there's a big opportunity here to engage in more social listening than thinking about social media as a publishing channel. Talk me through a little of the social media listening best practices.
4:57 Speaker 05 What a brand really needs to do is kind of understand the important conversations to them. So I talked about buying signals. Different brands will have different buying signals. It might be someone asking questions. It might be someone asking specifically about a product type. It might be someone engaging in a conversation that's positive with your brand. And then it's about, once you've found those bits of content, it's about leaning in to those particular conversations. And then providing information to the client, providing information to the customer and making that happen.
5:30 Speaker 04 It seems like a big challenge here when you're talking about social listening. You're essentially mining the social media channels. And instead of trying to push content out and start conversations, you're looking for where conversations are happening and trying to engage them. And hopefully they're related to your brands. How can marketers leverage sentiment analysis to convert social media insights and those sort of offline? They're not offline, but conversations that are about your brand, but not specifically tagging your brand. And how can you turn them into interactions that are personalized at scale?
6:02 Speaker 05 This is where AI really takes us into a new domain where we haven't been before. Cause it allows us to have individualized conversations automated at scale. That perhaps we have humans in the loop that are actually posting them. I don't think anyone's confident enough in the technology to be letting it just go free rein and automate. But it means that social media managers can have 80% of that content created for them. They can tweak it and then send. It out to the customers. The other thing you can do at scale is create insight. So we've had clients like bar chains that might have hundreds of different bars and restaurants. And they want to know sentiment across different branches in different geographic jurisdictions. What are people saying in this state or this country or this continent?
6:47 Speaker 05 And they're able to understand that voice of the customer of what's being said across those brands. I think that's really important to get that feedback loop. We're talking about how publishing is half of the battle. If you get the feedback loop, then the next time you publish something, that becomes really relevant to the people you're talking to. And you're not just making up content in the back room and you don't really know what's landing with your customers.
7:10 Speaker 04 Yeah, about 15 years ago, I started my first startup. It's called strum school.com. I think the website, I know I own the domain. I'm not even sure if the website still works. It was a guitar lesson marketplace to connect students and teachers. And I was thinking through the marketing strategy and at the time I was still working a full time job at eBay. And the head of public relations at eBay at the time gave me some advice. He said, you don't want to go and build an organic growth channel. You want to mine the organic growth channels for the conversations that are happening. And the example he gave me is instead of you pushing strum school.com can teach you how to play like Jimi Hendrix. You want to just look on Twitter for the people that are saying the words how to guitar. I will learn how to play guitar like Jimi Hendrix.
7:54 Speaker 04 And then figure out who is having those conversations and then respond directly to them. And so you're starting that personalized engagement and that essentially was the first sort of foray for me into sentiment analysis and understanding where the conversations were happening. You mentioned AI is making that easier. Is it about finding who is talking about things related to your brand? Or is it purely responding to comments with what we think is appropriate?
8:20 Speaker 05 It's definitely a mix of that. We use a technique out of Facebook Labs called Retrieval Augmented Generation. And we do two things with that. We look at what the client thinks is an important conversation and we try and find those conversations either on their owned brands or in conversations going outside their brands. When they find the places where those conversations happening, they need to then come up with a strategy of how they respond to that. It might be that they step into the conversation. It might be they go, so many of our audience is over here. We need to participate in that somehow. It's not enough just to reply. Maybe we do a sponsorship partnership. Maybe we partner with that channel from an influencer perspective. So they're kind of different
9:04 Speaker 05 ways to step into that conversation than just jumping in and replying. So we've just partnered with NWSL, the National Women's Soccer League. And obviously, as you can understand, they've got pretty chunky channels with lots of followers, lots of fans. And their big thing is around fan engagement. If they create lots of fan engagement during the week, during the build up, they'll sell more tickets, more people will come, more people will watch the game. So for them, it's about stepping into those spaces either on their channels or just outside their channels and making that conversation exciting, understanding where people are at and what they're looking forward to, what they're reviewing about last week's game and kind of making that really engaging, exciting for those fans, which is a mix of both of those places you were talking about. I've got to step into both of those.
9:49 Speaker 04 What was the name of the tool that you used from Facebook Labs?
9:52 Speaker 05 So it's a technique called Retrieval Augmented Generation. And this is where you take a brand's content of their own. So we might take their brand guidelines, their website, frequently asked questions, typically at a customer experience environment, you've got a cheat sheet that the agents use to answer questions. We take all of that. For a Canadian bank we did some work for, they had 130 pages. They get asked questions like, what's the interest rate on your credit card? And what's your mortgage rate? And things like that. So you take all that information, you team that up with a large language model that speaks English very well. And then that gives a response. So you put in the comment from social media and that gives a response in the brand voice of the brand themselves.
10:36 Speaker 05 And it's factually correct and it's tonally correct as well. And those two things are really important for brands.
10:42 Speaker 04 Yeah, that supports the copy creation, but the piece that I'm missing is how do you find the watering hole? How do you know where people are having the conversations about your brand that are relevant that you should respond to?
10:54 Speaker 05 It's a mix. So you need a listening tool, and that can be a listening tool on your own channels. So you kind of need to be analyzing those. And you need a tool that looks at the wider social media sphere to try and get some intelligence about where things are going on. The very simple way of doing it is with prompt scripting. So you write a prompt, which is, I am a retail brand called whatever my name is, I'm selling, you could try this on chat GPT, selling this water in the state of whatever, Ohio, and I'm looking to engage people that are going out to eating Italian food, for example. I'm making up as I go along. But what we're doing there is that's what we're looking for. And then
11:35 Speaker 05 what chat GPT will do is then help you with the content provided, help you target the right people. So you'll have a mass of social media comments, and then it'll find you the comments that meet your prompt guidelines, if that makes sense.
11:49 Speaker 04 Yeah, so you're using some of the platform LLMs, GPT, Claude, Gemini, to essentially do the legwork of scanning whatever social media networks you think are interesting, and finding the relevant comments. So you have some stuff to respond to. That's helping you identify your watering holes.
12:07 Speaker 05 Absolutely, yeah. There are some collection nuances that you have to have access to your own data, your own comments. So there's some work you have to do to kind of go grab them. But once you grab them, you can use those large language models to overlay onto that data and find you. We call them the golden needles in your haystack, your comment mountains, your haystacks.
12:27 Speaker 04 Here's the thing that I guess occurs to me, or even potentially concerns me. You can use the platforms to find the watering holes. You can use your Facebook lab tools to craft your responses, and you could essentially automate social media engagement in your brand's tone. But your brand reputation is more fragile than ever. So what AI-powered strategies should companies use to manage negative content? How do you avoid some of the sort of backlash and negative connotations?
12:58 Speaker 05 I always talk about criticism versus toxicity. And there's a number of things you can say. Most brands are entitled to remove toxicity off their channel. And by toxicity, I mean downright abuse. That is kind of not civil conversation, not safe for work, stuff that you don't want your children to look at. So what we advise our clients is you want a clear social media community policy. You want to put out there and you want to be really transparent. These are the types of conversations we encourage that we want. And these are the conversations that we don't want. So a news channel might be very open to all conversations. But for example, we work with a girl pop band and they don't want to have conversations about politics. They want to have conversations about music, which is completely fair.
13:42 Speaker 05 So they make that statement on their channel. They make that statement in the community policy. If you make comments about controversial political events, we're going to remove them off our channel. Because this isn't the space for that. There's loads of spaces you can go and have that conversation. We're not having it here. That's really where people can draw a nice line. People can understand what they're doing. So that's removing toxicity and that's removing what we would call unwanted content. And we would probably put spam in that bucket as well. There's a lot of spam out there that people wouldn't want on their channels. And then criticism is really do everything I've said before. Criticism, you should step into that. You should accept the criticism, have empathy for the client's experience with your brand or with your product or with your service and go like, I'm really sorry what's going on.
14:27 Speaker 05 What we found, so we work with Hinge, the dating app. And one of the things we worked on with them was if they just reply to review comments, they get an uptick of, I think, 0.9 on the app store by just responding to reviews. They don't even have to be good responses. They just have to show a bit of empathy. I'm really sorry for your experience. Some of them were like, we can't really help you with that. It's a user issue rather than an app issue.
14:52 Speaker 04 We just made the connection. If you mess up the date, it's not our fault.
14:55 Speaker 05 Yeah, absolutely. And what they find is they up their app store score by 0.9 as a result of stepping into every conversation, every review and replying to it. So that's how you deal with criticism.
15:07 Speaker 04 I gotta say, I'm totally shocked because that's a great example of somebody had negative experience, you want to reply with empathy, but my feeling is using the dating app metaphor, hey, you had a bad first date, so you're going to blame the dating app for the matching, when in reality, you haven't brushed your teeth in three weeks. And that's really the fundamental problem. So by us interjecting and saying, we're sorry it didn't work out for you, I would assume that that cultivates more negative responses, like you're engaging in a negative scenario. It's surprising to me that you're saying, actually, that provides a positive uptick. Explain the psychology there, because my feeling is like, side busting into a bad date would be a bad idea for a brand.
15:50 Speaker 05 I think the scenario you're talking about is probably less prevalent on those particular review channels, because people are saying, review the app. Some are people saying, yeah, I had a great day. Most people are saying, I downloaded your app and it didn't work on this. So there's a lot of kind of technical review problems people come up with. I paid for your subscription for three months. I had four bad dates, that type of thing. But I totally take your point. There is a line where you get to where people are just being downright argumentative, and you probably don't want to engage in those conversations. So there is a bit of a sense check on some of it. But prevailing is people want to be engaged with, they want to be spoken to. And if I make a complaint, someone listening to my complaint and saying to me, look, I've heard your complaint, I can't do anything about it. I'm really, really sorry. That evokes, at least, thanks for listening to me.
16:34 Speaker 04 Yeah, I think there's definitely some nuance in how that's managed, which is more reason why I think there should be human in the loop and not just using AI to respond in your, your tone and style guide. Using the dating app metaphor, hey, I had four bad dates after downloading this app, this thing garbage. And like, AI might respond and be like, well, we set you up on four dates, maybe hit the gym.
16:57 Speaker 05 And you're 100%. We find with lawyers, the legal teams in most brands, they are not legally ready for automated AI. And if we're honest, we are maybe two years into this technology being used relatively mainstream. I would say we're another two, three years away from having the confidence in certain low risk use Cases. So maybe not talking about how someone's physique needs toning up to go to the gym. How we're sort of maybe two or three years away from the tech really being able to kind of empathize in those situations. There's so many different types of AI. There's a guy in Australia who's raising a baby AI like literally from birth. I think the AI is now eight, nine years old. So his intention is to grow up an adult in the machine.
17:41 Speaker 05 That is the AI of Terminator two's police robot that Arnie put to death. We're a long way from that. This is AI being used for task oriented problems and doing very specific tasks really, really well. You've got to be very careful when you get into nuanced world that overlap three or four of those different tasks. And you ask it to be clever. I think we're two or three years away from that happening. And I think we're a long way from our policeman in Terminator two.
18:08 Speaker 04 As much as I love making dating app jokes, let's flip the conversation to be a little bit more positive. And talk about how brands are using AI to identify and amplify positive social signals that traditional media and marketing is missing. How can you use AI to find the good stuff and then syndicate it?
18:29 Speaker 05 It's exactly the same way as we've just talked about finding the negative stuff. And what we find is if you get rid of the negative stuff, actually, I use the Roadhouse analogy when I kind of demo our products. And I talk about, we're Patrick Swayze, we're the bouncer looking after the bar. And I know there's been a remake of that movie, but I've been doing this demo for like four years. So I go with Patrick Swayze.
18:50 Speaker 04 You gotta give an overview of Roadhouse for anybody that's probably millennial and younger. Give us some Patrick Swayze for us.
18:58 Speaker 05 Yeah, yeah. It just came out on Netflix again. But basically I've got a bar. I've got the lowlife are coming into my bar. They're smashing it up. They're throwing bottles at the band. And I need someone to come in and clean up my bar and get rid of this lowlife. So I can start cooking food and selling food and having a nice hospitable environment. So I hire Patrick Swayze to come in and clean up my bar. So we're that, we come in, we find the troublemakers, we kick them out. And once you've got that, what you find is. The bar attracts, you know, human nature is I want to go out and have a good time. I don't want to be beaten up. So I go to the bar where I can have a good time. And low and behold, that's not the one where the troublemakers are. So the whole point is you clean up your social media.
19:41 Speaker 05 You get a nice civil conversation going on or the conversation you want. And happy things start to happen. People have happy conversations. You don't really need to do anything. You just have to keep the environment calm and peaceful and happy and nudge that conversation in the right direction. And then as the add on where the AI helps you is you might be certain things that you do want to encourage because they're very specific to your brand and they're the sort of things that you want to kind of lean into. And they're the things you want to promote. So then you can use AI to find those things. And then you can then lean into those conversations and you might have influencer relationships. You might have specific products that kind of join into those conversations, which is why you're targeting on them.
20:23 Speaker 04 Talk to me a little more about the how. Great. I've Patrick Swayze and I've kicked out all the negative comments. It's just attractive people in the bar. And now, hey, everybody's good looking, everybody's having a good time. How do I tell everybody the pretty people are at my bar? How do I syndicate the message using artificial intelligence?
20:43 Speaker 05 Ads are a specialist topic and performance media directors at brands are constantly trying to crack the algorithms. Unfortunately, I don't have the answers to which algorithm sends you down the right path. What I do know is certainly on platforms like Meta, we have found a distinct change in the last two years. There used to be a rhetoric that bad stuff creates noise. So put loads of controversial stuff out there and you'll get noticed. What we have found is that Meta is now promoting positivity. So actually positive messages are getting higher up the stack. You're seeing that replicated in the functionality they're releasing. So they're promoting in threads, kind of positive engagement is going to the top of threads.
21:27 Speaker 05 So if you can encourage those things on your channel, then the algorithms behind certainly matter. Instagram and Facebook, and I think TikTok are generally following suit. X.com is. A bit controversial, they've sort of doing nothing. They're still kind of back where they were when Elon took over a couple of years ago. They haven't really moved the dial, partly because there's no one there to do that anymore. They all got laid off. So really Metas got a bit of a charge and TikTok a bit of charge of creating that positive feedback loop and building that into their algorithm. So the old adage of all news is great news actually isn't the case anymore on social media. Like positive news is getting promoted, positive engagement is getting promoted, and we're seeing real strong data that supports those hypotheses.
22:09 Speaker 05 We obviously don't know the exact way these things are written. We're guessing like the rest of the market, but that's what we're seeing.
22:15 Speaker 04 Social media putting positivity in the world. That's definitely something new. All right. And that wraps up this episode of the Martech Podcast. Thanks to Matthew McGrory, the co-founder and CEO of Arwen for joining us. If you'd like to contact Matthew, you can find a link to his LinkedIn profile in our show notes on martechpod.com, or you could visit his company's website. Again, it's arwen.ai, A R W E N dot A I. If you haven't subscribed yet and you want a daily stream of marketing and technology knowledge in your podcast feed, hit the subscribe button in your podcast app or on YouTube and we'll be back in your feed every week. All right. That's it for today. But until next time, my advice is to just focus on keeping your customers happy.
23:22 Speaker 00 Thanks for listening to the Martech Podcast and I Hear Everything production. Looking to launch or scale a podcast like this one for your brand, then visit ihereverything.com.
Previous Finding Your Niche Market Next 2025 social media landscape for brands trying to implement AI Page updated November 7, 2026 · transcript and summary are machine-generated