In this episode of The Sonic Truth, we dive into the evolving world of AI-driven audio advertising with Stas Tushinskiy, CEO & Co-Founder of Instrumatic. Stas shares how AI is transforming the way brands personalize, optimize, and scale their audio campaigns—unlocking new levels of contextual relevance and engagement. If you’re curious about how AI is redefining the audio marketing landscape, this is a conversation you won’t want to miss!
Listen to all episodes on your favorite podcast platform:
Transcript (Download)
Kristin Charron (00:01.635)
Welcome back to the Sonic Truth. Today I’m joined by Stas Tuschinski, CEO and co-founder at Instrumatic. Welcome Stas.
Stas (00:11.214)
Hi Kristen, thanks for having me.
Kristin Charron (00:13.327)
Thanks for being on. So let’s just, I guess, start with a quick primer on Instrumatic for any of our listeners that Instrumatic might be new to. So I have here the typical line of a voice AI marketing company enabling contextual video and audio ads. How would you just kind of, what’s your layman’s water cooler pitch, if you will, for those unfamiliar.
Stas (00:38.638)
Yeah, so we started instrumentic as a audio programmatic ad tech platform, helping publishers, audio publishers monetize their content with audio ads. And as we traveled our path, we’ve learned that personalization and contextualization is an effective way to increase performance of any kind of advertising, specifically audio. And so we started investing a lot of time and resources in developing multiple AI driven features.
which now enables our advertisers to go either from zero creatives to one or from zero to as many as needed in a matter of seconds, minutes.
Kristin Charron (01:21.615)
Wow, so there you could basically bring one creative, two instruments and spin that out courtesy of AI into a mouth.
Stas (01:30.264)
Correct, it’ll be the same voice, same music track. Everything stays the same, or maybe everything changes. It depends on the campaign goals and the advertiser needs. But the idea is we can optimize your creative essentially before we start the campaign by contextualizing it or personalizing it. So a good example would be mentioning the city where the ad is going to be delivered or the publisher name.
for delivering a unique promo code, mentioning specific customer offers like sign up bonuses or prices for air tickets, and actually mentioning travel destinations like flight from Europe Chicago for $3.99. And everything happens automatically and extremely fast.
Kristin Charron (02:15.429)
That’s awesome. just hyper-targeted, hyper-personalized. You mentioned this being powered by AI. How do you see AI evolving the way that brands engage with audiences through audio and podcast ads and kind of all applications of audio, both this year and moving forward?
Stas (02:35.658)
there are so many directions where the tech will go in terms of applying AI to audio advertising. So let’s break it into kind of separate parts. So the first one, which I just touched on, is the ability to personalize every single audio ad campaign that brands run. Because if you think about it, if you could personalize, why wouldn’t you? And so effectively,
Sorry, eventually we expect that every brand will personalize their ads or contextualize, or at least do some kind of A-B testing where it’s not about making ads personal, but you can just try different calls to actions or different intros, different voices, accents, a lot of stuff you can try and A-B test. Then, and that’s what we are working on, AI will start learning on…
based on previous campaign performance to suggest better ways for optimization of future campaigns. And it won’t be just a blind guess. It will be based on the historical data broken down by brand category, advertiser, time of the year, time of the day, and many other factors. Then you mentioned podcasting. That’s another big industry that’s about to be
upgraded by AI, you will. And the idea is now AI can clone voices. And in many cases, you can’t tell the difference. And very soon it’ll be like close to a hundred percent accuracy all the time. So factually what that’ll enable for podcasters is the ability to sell their native host read ads programmatically. So an advertiser would bid for an impression.
then a platform like Instrumatics will clone the voice and use the podcasters voice to deliver the brand’s ad in the host voice to make it native. And everything will be done automatically. Now what that means for the showrunners is significantly greater fill rates and CPMs because they’ll be able to offer targeting, which can dramatically improve their ad revenue.
Kristin Charron (05:01.883)
would you say to brands and showrunners that are intimidated by AI and all that it can do?
Stas (05:09.546)
Just like any other tech, the sooner you embrace it, the better you would feel. Because, I mean, it’s inevitable that all new technology, not just AI, would eventually become something that everybody uses by default. And so I think the sooner we embrace it, the sooner we learn how to use it appropriately. And the sooner we learn what works with doesn’t, the better our job
Kristin Charron (05:15.099)
That’s fair.
Kristin Charron (05:41.797)
For sure. And you mentioned AI being able to, brands and showrunners being able to use AI to transform post-red ads. What role does it play conversationally in shaping more personalized ad experiences?
Stas (06:03.693)
you mean like replicating the host style of the ad delivery or something else?
Kristin Charron (06:10.763)
Yeah, like you leveraging AI from a conversational perspective to, you know, write the ads and make sure they’re contextual.
Stas (06:17.527)
and
Yeah, for machine it’s rather easy if you can provide it with, let’s say, a number of examples how that particular podcaster did it previously. And so let’s say you upload like 100 examples, or if there are more than even better, it’ll quickly catch the style of the host, and it can replicate quite easily.
Kristin Charron (06:48.769)
Okay. And then, from your perspective, what are the biggest challenges that brands face when integrating AI into their marketing strategies and do have any tips for overcoming them?
Stas (07:02.038)
It’s a great question because the biggest challenge, it’s kind of funny that the biggest challenge is the organizational legacy, if you will. So the way the process were set up for the 20th century advertising placement, meaning everything starts at the creative agency level where you…
come up with ideas, brand guidelines, then you produce a creative, then you send it to a media team that’s supposed to place it across various media channels. And now what AI enables us is real-time creative optimization, where it’s not changing how the brand sounds or looks, but it can just tweak the creative, try, you know, A-B test it, try different approaches, So effectively playing with the…
Kristin Charron (07:37.019)
Mm-hmm.
Stas (07:56.494)
assets that were provided by the creative agency. And this is where the industries is facing a challenge of approvals. It’s all of a sudden, you have to approve those changes close to real time. And you already kind of hand it over to the media team. And typical media team, they don’t want to touch the creative. Most of the times, they’re not allowed to touch the creative. And so it has to go back to the creative team.
Kristin Charron (08:11.673)
Right.
Stas (08:24.712)
And that just creates a lot of friction. And so essentially, and a lot of people are realizing that they need to upgrade or update their business process to actually, well, to adapt to the new era of advertising, which is machine-driven creative optimization.
Kristin Charron (08:35.508)
Mm-hmm.
Kristin Charron (08:45.663)
So in that sense, and excuse the naive question, but I’m not close to this. How do they adapt it? it, do they bypass approvals? Do they trust what the machine is tweaking, if you will, or how do they get, how have you seen them get close to real time approvals?
Stas (09:05.42)
I think eventually you’ll have a person, maybe a new role on the advertiser side, that would have enough authority to approve changes to the creative. Again, we’re not talking about producing completely new creatives or something that no one has ever seen before.
Kristin Charron (09:18.459)
Mm-hmm.
Stas (09:26.71)
and were not approved. So we’re talking about kind of creatives following the brand’s guidelines. So there has to be a person, not like the entire creative agency team, but just one person who is authorized to approve those small changes here and there, almost like in real time. Because it’s not like we’re changing the machine is changing the entire storyline, the entire brand positioning. It’s just like,
Why don’t we change three words here, five words there? If we’re talking about video ads, why don’t we try a different scene sequence? And so there has to be somebody who can give it a go immediately. Right, if you think, exactly, and if you think social media, that requires even faster reaction. you, I mean, imagine brands incorporating user-generated content into their communication.
Kristin Charron (10:04.891)
Yeah, it’s really almost like a quality check at that point since it’s not really a material change.
Stas (10:22.796)
using it.
Kristin Charron (10:22.843)
Very true. Yep. Yeah. How does AI help measure and improve the efficacy of audio campaigns in real time? Are you seeing any changes there?
Stas (10:25.612)
Yeah.
Stas (10:38.11)
absolutely. So, we, by we, I mean, very tonic, instrumentic and, Densu we did, I think a fabulous case study last year, which by the way, won the ad exchange award. so we, we did a campaign for Intel together with Densu. and we produced many versions for them.
Kristin Charron (10:56.271)
Yes, it did.
Stas (11:06.35)
they used three context parameters in that campaign. One was they mentioned the publisher at the beginning of the ad to grab more attention from the user. So it was like, know Spotify listeners, like, appreciate good tech, Pandora listeners, and so on and so forth. And that was done to get more attention from the user. Then the second parameter was the countdown to Black Friday.
like with seven days left, with six days left, and so on and so forth. And the third parameter was the retailer where you can buy an Intel-enabled laptop, like Amazon and others. And so we did a case study and Veritonic executed the research. And the results were actually amazing. One of the major takeaways was that consumers can’t tell the difference anymore between an AI-generated ad
and a human-produced app. Because we saw that even an AI-generated generic app outperformed
human-made it. And we saw that across other brands and campaigns that we ran. But then the personalized versions, they took those gains even further. And brand metrics increased between 12 and 22 percentage points across the board. So what this tells us is that AI can dramatically improve campaign performance, even in audio. When I say even in audio, I’m obviously referring to the fact that
Audio just like CTV has the click problem, just like there is nowhere to click. So it’s really hard to measure the last mile. so, you know, case studies and the research that we did together was really important. And it showed that obviously personalization works, but AI driven personalization optimization works as well.
Kristin Charron (12:50.009)
Right.
Kristin Charron (13:06.171)
So it’s again, just making the case that AI is not the enemy. It’s something that can really help brands and marketers and help them streamline things and still be effective.
Stas (13:21.346)
Yes.
Kristin Charron (13:23.771)
Do you get the sense that, I mean, that case study was fascinating, obviously, selfishly, but in general, do you get the sense that brands are looking for more of that sort of proof point that AI is something they should embrace? And if so, what is Instrumatic doing to kind of help with that message?
Stas (13:50.35)
A lot has changed over the last year. So a year ago, we had to have such conversations about like applying AI in general. Should we do it? Should we not do it? Is it a threat? Is it an opportunity? Now, I can’t remember having these conversations for a number of months now. Now the real question that everybody is asking themselves, how should we do it?
given our own organizational setup, goals, and all that. But how should we do it? Who can we trust? Who should we bet on? Because clearly, the tech and the industry, I’m referring to the AI industry right now, is changing so fast. mean, everybody wants to integrate a solution or a number of solutions and don’t have to switch them something else a year later. So there are many considerations, but clearly, everybody who I speak to right now
Kristin Charron (14:18.085)
Thank
Stas (14:47.854)
they don’t have this question anymore, which is, should I try it? The question that they have, how can I do it, and with whom?
Kristin Charron (14:56.249)
Yeah, that’s super telling. As AI capabilities advance, what ethical considerations should brands keep in mind while leveraging this tech for audio marketing in particular?
Stas (15:09.13)
I think it should be the same ethical considerations that we had previously. AI is just a tool. It’s not something that replaces humans. And it’s not really great at making decisions. It’s really great at suggesting multiple or maybe actually limitless ways to improve what we do. But effectively,
it’s up to a human to make the final decision. Like for example, if we’re talking about audio ads, then a AI can suggest, for example, better copy. It can suggest different voices, but it can’t tell you which version resonates better emotionally. Because there is something in the emotional aspect that is a little bit hard to measure with math.
It’s just like art. Sometimes you look at something that looks weird, but you feel something when you look at it. And so, Yiayi is not there yet. Maybe it won’t be ever there to understand women’s emotion.
Kristin Charron (16:12.805)
Well, that’s what Veratomic is for,
Stas (16:16.748)
Yeah, exactly.
Kristin Charron (16:19.501)
And in your conversations around AI, does brand safety ever come up in terms of like its capabilities or lack thereof? And I think this, this kind of speaks to your latest point as well. Like that human touch, that decision making will always need to sit with, you know, a real person that sees the full context. But how often does brand safety come up in these conversations, if at all?
Stas (16:46.862)
Yeah, it comes up quite often.
There is this fear that some people have that AI will get out of control. mean, effectively when we talk about advertising ads, they represent brands. And a bad ad campaign can dramatically hurt a company. We all know multiple examples when everything went wrong with, you know, after a wrong ad was placed. So ads are important because they represent the brand’s values and everything else. And so…
Clearly, agencies and brands, want to have full control over the final result. And we put a lot of limitations on AI. So before we allow our platform, our tech, to produce any ads or start serving them, we always get approval from the brand. We will show these are the limitations. These parts of the ad.
will won’t change and you know, some other parts will change. And after, you know, after the advertiser proves that we go into mass production mode and the campaign goes live.
Kristin Charron (18:02.991)
Very cool. And do you have any predictions as to how the advancement of AI and its role within audio advertising in particular may evolve over the next even year or so?
Stas (18:22.762)
I think the short answer is scale. Everything is going to reach a normal scale. if last year we were talking about 20, 30 versions per campaign, right now we are serving campaigns that have like 500 versions of an app.
so I think a year, year from now we’ll be talking about a lot more version and it’s not just like one time production, but also refreshing creatives as the campaign, is being served, then creating a series of ads where, you know, when the user gets, let’s say a second or third contact with the brand, the user would get another ad that would kind of continue the story, give another reason to buy the product that the brand is selling.
Another very interesting use case is going to be remarketing. Audio ads, think, would be phenomenal for remarketing because when you think about remarketing, it’s not about like…
It’s not about searching or finding for new buyers. It’s about reminding your old customer who’s not your current customer, why they should come back. Give them a reason, give them a promo code, give them something, or just simply reminding them that, you know, we had a good experience together like six months ago. And so audio ads would be phenomenally great at it because…
Kristin Charron (19:44.635)
Sure.
Stas (19:49.486)
There is no need to drive installs and nothing like that. The consumer rating knows about the company or they have the app or whatever. So they know where to buy stuff. They just need to be reminded of that business and maybe some new developments, any products that they’re releasing.
Kristin Charron (20:07.279)
And you talk about the scale that AI enables in terms of creating many iterations of ads. I have to believe that there’s a cost savings there and that the brands and advertisers don’t have to spend as much as they would if they had to create multiple iterations themselves. Do you have any insight or thought as to where the budget that they’re saving from AI is likely to land, where those dollars might shift to?
Stas (20:37.678)
more media buying.
Stas (20:43.252)
If you spend less on creative production or creative optimization, then either you just save money on marketing or you buy more media.
Kristin Charron (20:46.181)
Sure.
Kristin Charron (20:53.519)
buy more. Yeah, absolutely. Is there anything else that I know that kind of wrapped up the questions that we had spec is there anything else that you want marketers or brands listening to this to know about AI about audio advertising at large? How it could, you know, play a role in their strategy for the year ahead?
Stas (21:15.79)
Yeah, there is.
There’s something that I think marketers should think a little bit more about and it is omni-channel marketing. So what if the user journey in terms of being contacted with the brand’s ads consists of multiple steps across multiple channels. So let’s say I’m commuting and that’s the first time I’m hearing one of the ads from the brand and it’s audio. Then I’m at home watching TV and that’s my second contact.
And maybe then I go to, you know, then I open one of the social media platforms. It’s my third contact. And it would make a lot more sense for brands to think through the entire user journey. what, if my first contact is audio, how should it sound like? Or if my second contact is audio, what should I say? It can be the same generic ad that’s being served over and over. And all of the channels, you know, are being planned and…
and thoughts separately. Now we have the capability to tie it all together in one single storyline.
Kristin Charron (22:29.827)
Could not agree more. We talk about that a lot. It’s almost like, especially in audio and those different touch points, the sound, your, your brand is becoming part of the soundtrack of people’s lives and what they might want to hear to your point in the morning on their commute, when they’re on the subway, isn’t necessarily the same exact message that they’re going to want to hear when they’re listening on their smart speaker and cooking dinner, right? Like it’s contextual. It’s depends on their situation and what else they’re doing. And if they’re multitasking, I think that’s a really great point.
Stas (22:59.468)
Absolutely. the benefit of audio is you can produce ads really fast and you can respond to even some global news that just happened or some other events. There is no video component. So just getting those approved, even sometimes like there is no need to approve them because you heard the voice, you heard the ad, the construct and you know how the new one is going to sound like. So you can go give a green light without even
listening to the ad and that just creates this enormous opportunity to always be relevant for brands with audio. Like, I don’t know, there is like a sports event or something else and you can immediately change all your creatives and be relevant. And that just gives such a competitive advantage to the brand where they can continue the conversation with the audience that’s evolving over time. So it’s not like a three months.
campaign or six or even some brands run the same creative for 12 months. It’s like always evolving conversation or communication.
Kristin Charron (24:08.751)
Yeah, I couldn’t agree more. And one thing that stood out when you said that was the brands that are active on social during the Super Bowl in particular. You know, they’re engaging in real time. They’re talking about the commercial. it’s, it’s almost as if they’re watching the whole experience with you. And there are a few that still stand out that have done things over the years that I still remember, even though I’m not necessarily engaging with that brand every day, it was impactful. And so I think definitely the same goes for audio and for the use of AI within the process as well.
Kristin Charron (24:41.335)
Excellent. Well, thank you so much, Stas. This is such a pleasure. Definitely want to keep in touch with you and keep up with all the latest and greatest things that Instrumatic are doing. If our listeners are interested in learning more and asking any questions about your process and services, where should they visit?
Stas (25:02.062)
So we’re active across all major social networks as a company. just search for instrumentic. I’m also active on LinkedIn, so you can follow my stuff.
