Humans in the Loop — Why AI on the Supply Side Still Needs Media Experts | Behind the Numbers

In today’s podcast episode, we discuss why supply-side optimization is happening now, how AI and human judgment should work together across programmatic, and what buyers should be demanding from their supply partners.

Join Senior Director of Podcasts and host Marcus Johnson, Principal Analyst Yory Wurmser, and Vice President of Enterprise Partnerships at AI Digital Josefina Schmella for the discussion. Listen wherever you get your podcasts, or watch on YouTube or Spotify.

Subscribe to the “Behind the Numbers” podcast on Apple Podcasts, Spotify, Pandora, Stitcher, YouTube, Podbean or wherever you listen to podcasts. Follow us on Instagram.

Smart Supply by AI Digital is the supply intelligence layer built for programmatic buyers. Powered by AI and 450+ media experts, it scores and gates every bid request across 12 SSPs before your DSP bids, so inventory arrives optimized for your KPI, with every AI decision visible. Learn more at aidigital.com.

Episode Transcript:

Marcus Johnson: [00:00:00] Behind every bid is a supply decision most buyers never see. That's why they turn to Smart Supply. AI Digital Supply Intelligence solutions scoring and gating every path before the DSP bids, so inventory arrives optimized for your KPI. Learn more at aidigital.com.

Hey, gang. It's Monday, November 23rd. Yory, Josefina, and listeners, welcome to Behind the Numbers, an EMARKETER podcast made possible by AI Digital. I'm Marcus, and joining me for today's conversation we have, living in New Jersey, EMARKETER principal analyst who leads our advertising, media, and tech desks, Yuri Wormser.

Yory Wurmser: Hey, Marcus. How are you?

Marcus Johnson: Hey, fella. Very good. Thank you for joining the show. We also have, calling California home, VP of enterprise partnerships at AI Digital, Josefina Schmella. Hello.

Josefina Schmella: Hi, Marcus. How are you? Good to see you

Marcus Johnson: both. Welcome. Good to see you. Good to see you. Anytime we have a special guest on, we of course start with a speed intro.[00:01:00]

So first question is just for you, Josefina. What does your company do in a sentence, and what do you do for them in another?

Josefina Schmella: AI Digital, we're the orchestration layer within the programmatic ecosystem. So we provide the supply intelligence to enable agencies and brands to more effectively reach their programmatic outcomes.

Brilliant. What I do is I'm the vice president of enterprise partnerships. Within Smart Supply, I negotiate our contracts with our holding company and large agency clients.

Marcus Johnson: Excellent. And second question is for both of you. Josefina, you can start. What's the most interesting thing you've read, watched, or listened to recently?

Josefina Schmella: Okay, so I'm a mom of three. I don't have a lot of spare time so I tend to multitask when I read, meaning I Audible a lot. I'm in a book club, and I would say Project Hail Mary is the one that I fell in love with. Oh. That book, when I would walk, it would just suck me in. Yeah. If you haven't read it the two characters in it, Dr.

Grace brilliant [00:02:00] scientist but he's also scared of a lot of things, and he, he- the Iridian force, which is essentially Rocky, the alien that you get to meet in the book, both of them have depth to their character. They're brilliant. They're scientists trying to figure out how to save Earth in the story.

And so I fell in love with them as a character, and I think also just fascination around science and math. Those were my favorite subjects growing up.

Marcus Johnson: Yeah.

Josefina Schmella: So it kinda took me back when they're critically thinking and solving some of the problems. So Project Hail Mary, highly recommend it.

Marcus Johnson: Great choice.

Saw the movie, haven't read the book, but the movie was brilliant. And I want every T-shirt that Ryan Gosling wears in that film, 'cause they're all funny. Yuri, how about you?

Yory Wurmser: It's a great book and I thought it was a really great film adaptation as well Project Hail Mary. I'm not gonna say you stole that from me because I wasn't gonna say it-

but that, that's a great one.

Marcus Johnson: Unbelievable ...

Yory Wurmser: I was gonna go on a interesting/silly angle, which is just Cunk on Cinema. It is very funny- Oh ... and very,

Marcus Johnson: Yes. Yeah. She

Yory Wurmser: is great. Very interesting.

Marcus Johnson: [00:03:00] Yeah. Yeah, it's a mockumentary style that she's done. She's done a bunch of TV shows.

Anne Droid, have you seen that, Yuri? No. She's where she plays, yeah it's this British, comedian, presenter, writer producer, but she does a show in the UK where she pretends to be a robot, and it's a world where robots are now humanoid robots are helping out in the home. And she plays it brilliantly.

But yeah that's a good one. I haven't seen that one yet, but I'll definitely be watching it. Excellent. There are our two guests for today's episode, but today's real topic is humans in the loop, why AI on the supply side still needs media experts.

All right. Let's just dive in. Josefina, first one straight to you. How did 15 years of innovation on the demand side, platform side leave the supply side behind, and what limitations has this created?

Josefina Schmella: Yeah, so h- as you mentioned for the last 15 [00:04:00] years in programmatic, the innovation has existed on the demand side.

The DSPs are the ones that created the algorithms, the data, the measurement. Every- everything has lived there. Essentially, supply was something that we bought in programmatic, not something that we optimized, and in the last 12 months, that's changed. I think there's, two universes which y- most of the industry knows in programmatic, the open exchange and the private marketplaces.

Each one had different pros and cons. Private marketplaces had the benefit of quality, control, transparency, but n- there's a cap to scale, and also not as efficient, where open exchange had reach and efficiency, however, less control and brand safety. And what's happened in the last 12 months is that the innovation has accelerated on the supply side of programmatic, and it's a really exciting time where now we're able to optimize an essentially clean supply based on [00:05:00] historical supply signals and outcome signals- to serve that supply to the DSPs before they even bid on it. So it's a new, this advancement is accelerating now. I think historically the space was just, all the innovation was coming on the demand side, but new advancements that are coming in programmatic on supply have just happened on the, in the last 12 months, and that's what's allowing this to now come to the table.

And so there's m- more advancements actually coming in the next six to 12 months, which is really what's exciting, 'cause I've been in our industry for 20 years and I saw the innovation hitting programmatic on supply and I was like, "This is exactly where I wanna go. I wanna- ... be involved in this," and that's really the reason why I selected AI Digital to join the team.

Marcus Johnson: Yeah.

Yory Wurmser: Yeah. I would just add that I think everything that Josefina said is accurate, and I'd add that, header bidding, when it came around m- maybe 10 years ago or so, commoditized SSPs into supply side in a way where you could just, you'd [00:06:00] see several different, the same inventory from different companies different sup- supply side companies and that, gave, I think, the DSPs an advantage.

I think exactly what Josefina mentioned, you're starting to see that, that supply side differentiate itself a little bit by being closer, that they have intelligence that they can present that is differentiated, and has let them fight back and curate on their own side.

Josefina Schmella: I think it's fascinating in that the- DSPs are cutting out some of the SSPs.

The S- SSPs are leapfrogging some of the DSPs. There's companies that are on both sides of the table, and so there's, that's forcing a lot of inno- innovation and competition that's happening in the space. And I think it's propelling the industry forward right now of what is happening and what's unraveling- Yeah

Which is really exciting.

Marcus Johnson: So I want to rewind and then fast-forward. And so what I wanna do first is ask you why is the shift towards supply side optimization happening now? And then afterwards we'll talk [00:07:00] about what you're excited for the next six to 12.

Josefina Schmella: Okay, great. Yuri, you want me to go first?

Yory Wurmser: Yeah, might as well.

Josefina Schmella: So I think there's three main things. There's more variables in this, but the three things why this is happening now is, one is that the value is moving really fast. The amount- value's moving fast in that the DSPs are cutting out the SSPs. The SSPs are reaching past the DSPs and there's companies that are on both sides of the table, so it's forcing innovation to happen.

So that's one thing. A lot of the industry events I've been into recently, they're talking about how much innovation is coming into the supply side. So the data, the measurement the abundance of signals that are available today. So it's not that, do you have access to the signals, it's which signals are you paying attention to- and then how is your algorithm- Yep ... leveraging those signals. Second I'd say is there's new rails and technology advancements that are out, like ARTF, agen- [00:08:00] agentic real-time frameworks. Those are allowing supply to be optimized before it goes to the DSPs. So I think the technology advancements that's happened in 2026 are one of the things that are providing this opportunity, is the easiest way to think about it.

And the third is the signal is open. So a lot of the rich signals historically were reserved for direct partnership, and signals are more readily available. There is an abundance of signals that are now available in programmatic, so it's not about having all the signals, it's about how does your algorithm which signals is your algorithm using, and then how is it leveraging those signals to be most effective and impactful-

To clean that supply and optimize it.

Yory Wurmser: Yeah. And I think a, a common thread in a lot of this is AI just maturing. ARTF is a, is an agentic platform to basically bring a, a lot of this data and decisioning together in one place. And it works because it [00:09:00] happens p- very quickly with AI.

Everything decisioned very quickly and transmitted. And as you said, there's just, there are more signals and more meaningful signals, in part coming through first of all, AI interpretation, but also AI decisioning on what's important and what isn't. So those two, in, in a lot of different ways, AI is contributing to this.

Marcus Johnson: So actually before we look ahead, let me ask this instead. Josephine I'll start with you again for this one. What's the optimal relationship between how AI and human judgment should work together across the buying optimization planning all of this stuff? 'Cause I think that's probably been a big question for a lot of folks is I was reading a good article in Axios the other day and it was saying companies have adopted AI but they're still trying to figure out how to use it or what they're using it for.

They're like, "Here it is," and everyone's like, "Okay." What do I do with it?" Yeah, what should the optimal relationship be in between AI and humans in this regard?

Josefina Schmella: Yes. Within programmatic, so AI Digital has a headstart of nine years because the company has been around for nine years. And [00:10:00] what we'd say is agents guided by humans, meaning humans should be in the driver's seat.

We're the ones setting the objective, the goals, using our expertise, and then allowing AI to do the heavy lifting. AI is what should be providing, in programmatic, visible decisioning. So it should be what's saying, "Okay, these are the SSPs that need to be included. Here are all the supply paths, and I'm gonna score those for you so you see the quality of the supply paths."

It should be looking at brand safety, IVT, MFA, pricing, so CPMs. All of that should be surfaced by the AI as well as the historical data, and humans should be looking at that and saying, "Okay, what's the trade-off here? Great, I get more outcomes. What, how does that impact my CPM across all the different channels?

What does that mean for brand safety?" So AI should be surfacing it to give visible decisioning so humans can lean in and use- ... their expertise to actually set up those campaigns and be [00:11:00] clean.

Marcus Johnson: Yeah. Yuri, I love this idea of as Josephine just said, AI as the passenger, people as the driver, and they're pointing things out as you're going, giving you direction.

But it's the human that's in control of where the ultimate destination is.

Yory Wurmser: Yeah, that's exactly right. I, the AI... Humans are so important for the, the overarching strategy, for the goal setting. And then in terms of governance as well you need to ha- have oversight to make sure that you actually are pursuing the goals, the AI is pursuing the goals you want it to pursue.

And you just wanna have some degree of control and accountability as well. And that all boils down to still having a human in the loop.

Marcus Johnson: Let's circle back to, so Josephine, you were saying before about there's been a lot of supply-side optimization happening now, but you're excited for what's to come in the next six to 12 months.

A big reason yeah, why you were saying that AI Digital was important to you. W- what are you most excited about?

Josefina Schmella: I'm most excited, I think there's a couple things. One is we [00:12:00] talked about ARTF and containers. I think that's, we're just at the tip of the iceberg of what's in there. I think everyone's figuring out, okay I can put data in here I can put measurement in here, and case studies are coming out about why that is really impactful.

But as we know in our industry, you start to test the waters, and everyone gets smarter, AI plus people. So I'm excited about what that's gonna mean for the impact of media results and programmatic.

'Cause it's gonna make those media dollars be more effective, so I'm excited for that. I think I'm also really excited and this is one of the things that drives me, being in ad tech for 20 years, for me, I'm like, if we're delivering results for clients and agency partners and their businesses are growing then I'm working at the right company.

And I think that is part of the reason I joined AI Digital, is because I do believe in quality, cleanliness of programmatic supply, because as we all know, there is a lot of fraud that's been in our industry in programmatic. And so I think the advancements that are coming right now with [00:13:00] clean supply and everything that's possible today is gonna allow us to remove more of that and identify it faster- to allow investments to be more effective, and ultimately to drive more sales and conversions for the clients that are advertising in the ecosystem.

Marcus Johnson: Yeah.

Yory Wurmser: And I'd extend what you're saying a- and I agree with e- everything you mentioned, is just that with a- agentic media buying starting to become a reality, ARTF is an early...

not an early it's part of that it- agents interacting with each other on the supply side. But when you start linking buyer, buyer agents and seller agents, which you're probably gonna start seeing more and more over the next 12 to 18 months that- That's gonna be super exciting, and a lot of that innovation is happening on the supply side.

Some of the earliest kind of decisioning areas for that is happening, within a supply side realm. So I think it's a really interesting, maybe among the most interesting 12 to 18 months we've had in a long time- ... I think what's coming up.

Josefina Schmella: Yes. So that's [00:14:00] actually I left that out of my response, but Yuri, you just meant- you struck a chord with me, which was right now what's happened with how A- humans are using AI and the growth in 2026 quarter over quarter, it's impacting programmatic, and I think it's what's really exciting is there are a lot of agents that act like humans that are in the programmatic ecosystem.

They're looking, and they're not, it's not fraud, it's they're looking for information. And in 2027, us instead of, I think as people recognizing that is happening and actually starting to set strategies of, hey, all these agents are looking to research on behalf of people for what products they wanna buy, and comparison shopping, how do we give them the right data?

How do we influence their decision? Is honestly a whole new opportunity instead of, in addition to buying media, like that is another intentional proactive strategy that we should be thinking [00:15:00] through. And it's an exciting opportunity.

Yory Wurmser: Yeah, I just wonder how much consumers are gonna buy into that.

I agree. It's a sup- it's an exciting and super fast-moving area. I'm curious if people are gonna actually delegate a lot of their shopping to Muse and to ChatGPT, or if they're gonna use it as an aid. Regardless, it is still important to get your information out there for the agents.

But I, I talk about this with my colleagues here. It's something that's cha- gonna change shopping. It just, the question is, like, how much do the average, does the average shopper want to delegate to these agents? It's super fascinating.

Josefina Schmella: W- what is fascinating for me is that it's n- not just on the consumer side, but from the tech side.

And you're thinking about agencies and brands and programmatic buying the media, and these agents are out there looking to find that information. It's already happening, and it's not necessarily, that is already happening in the ecosystem, and so I think the more that on the ad tech side, as we recognize that [00:16:00] is something that's happening, like, how do we provide those agents with the right information to make sure the brands are being considered when people are, when these agents are looking for that information is important.

And I think consumers will catch up to some degree on what's happening. A- again, I don't know about adoption like you mentioned, Yuri, on all of that. But it's like this is happening underneath the surface already. Yeah. It is, those impressions are going to some of those agents in programmatic today.

And that's also why supply intelligence is really important, is because we can tell the difference between where the ad is gonna be served and what's happening in that supply path. That's something that can be filtered out. But tho- they are underneath the surface. They're looking for that information, so it's just a, for me, I'm like, what should our industry be intentionally doing to actually- Empower them.

Yeah. Because I think c- certain companies are doing it well, and there's a lot that are getting left behind right now.

Yory Wurmser: Yeah. I would say the majority are being left behind. J- Yes ... it's really, it's the unusual ones who are on top of it.

Marcus Johnson: Yeah. Let's [00:17:00] end with this, folks. What should buyers demand from their supply partners to ensure transparency, quality, efficiency, and performance?

Josefina?

Josefina Schmella: Yeah. I think they need to, one is they need to demand visible decisioning. So as the AI surfaces, here is the supply, here's brand safety, IVT and also pricing. They need to be able to see it, not just say we're providing you with supply intelligence," and it's like, "Trust us, we're gonna do it."

It's, "Let me see it. I wanna see the technology." So you should be able to see a platform or the tech in action. So that's one thing, I'd say visible decisioning. Two is the results, the outcomes. So this is not all j- just to be done for the sake of being done. It's like, how does this allow your media dollar to be more effective?

And part of that is when you're launching a campaign and looking at the results, is your dollar more effective? And that, that's something that should be demanded in there.

Everything else is, SPO, IVT, all the other stuff. Brand safety, that's included as well, but I would [00:18:00] say visible decisioning and outcomes are the two must-haves.

Marcus Johnson: Yeah.

Yory Wurmser: Yeah, and I, and they're interrelated with the idea, they're combined for accountability. If y- you need to be able to see what's happening, and you need to p- you need to be able to see what's working. And combined, that's just the accountability you'd expect from any provider.

Marcus Johnson: All right, folks. That's all we've got time for this episode. But let's finish off with some key takeaways. Number one, Josefina, something you've been talking about throughout, about agencies and brands unlocking greater efficiency through better supply, higher quality inventory, and AI that predicts which paths will deliver the best outcomes at the lowest cost this being the power of supply path optimization, SPO.

And then number two Yuri, you talking about AI driving significant innovation making it increasingly important for brands to determine how and where they should oversee, supervise, and guide AI development to ensure it aligns with their business objectives. But that's all we got time for this episode.

Thank you so much to my guests for today. Thank you first to Yuri.

Yory Wurmser: It's great being here.

Marcus Johnson: And [00:19:00] of course, to Josefina.

Josefina Schmella: Thank you for having me. It's great to be here.

Marcus Johnson: Thank you. We appreciate it. Thank you so much to the production crew Stu, who runs the team, Danny and Mike helping us out with this one.

Thanks to everyone for listening to Behind the Numbers, the marketing podcast made possible by AI Digital. On Wednesday, Suzy will be here on the Reimagined Retail show talking about the November Retail Awards, and then I'll be back after Thanksgiving on Monday the 30th, I believe it is, of November with more Behind the Numbers, talking about what the GTA VI release reveals about the state of gaming.

Get more articles - create your free account today!