From 5 Months to 15 Days: How Exact Agency Saves Campaigns and Lands Clients with Hedra

We sat down with Joni Dobrov, Chief Creative Officer at Exact Agency, an AI-native creative studio built at the intersection of cinematic production, brand storytelling, and modern technology based in LA and Nashville to hear her perspective on how AI is changing the marketing landscape and how Hedra has helped them land new business deals.

Some highlights that stood out were the examples she shared of how AI saved some of her clients marketing campaigns and expanded product offerings thanks to the quick iteration process.

Top Highlights

Eyewear Campaign Saved: A global eyewear company had a hard deadline and Exact was able to deliver a full campaign in just 15 days, a production timeline that would have taken four to five months through traditional means.

Candy Company Expanded Their Offerings Due to Fast Iteration: By streamlining the creative iteration process for a top Amazon confection brand, Exact Agency helped the client rapidly test variations and attract new demographics which ultimately expanded distribution into multiple grocery store chains nationwide.

AI-Powered RFPs Are Winning New Business: Exact Agency now uses AI video tools including Hedra to build immersive, brand-rich pitch decks and visual walkthroughs for RFPs, including 50th anniversary events and entertainment venue concepts in Chicago and LA saving clients time and budget.

"Soul Behind the Eyes", Why They Keep Returning to Hedra: Joni credits Hedra's micro-expression quality and character depth as the reason the team consistently comes back to it for brand and narrative work, describing the moment the uncanny valley faded as the creative breakthrough that defined their AI workflow.

Full Interview

What was the initial motivation for starting Exact Agency?

The company is very adamant about helping brands, clients and creators tell stories that otherwise without the help of AI, would be too expensive, too slow or too risky traditionally.
And as we continue to grow here, we're very adamant about letting people we work with know that we are human-first and very much human-led creatively. AI is just a tool and we're leading it.

How did you learn about Hedra?

It was about a year ago when we discovered Hedra and it started with a conversation with an intern that was on staff at the time. We were looking around to see the tools and the capabilities, because given at the time, no one tool rendered equally. I still believe that today.

But at the time, we had yet to discover micro-expressions as strong as those I witnessed in Hedra. It was a couple different pieces that we were iterating on that I was just so impressed by. We then naturally went into conversations with the engineers, and we were giving really cool tool assessments. Since then, we've been working very closely with the Hedra team, and we've had some really good camaraderie. We've almost made some friends. Hedra's been a very good support. They have a wonderful community, and we always go back to Hedra when we're working for brand and narrative.

Was there a specific moment early on where you thought “that okay that's the thing” to use for AI content creation?

So much of the market is flooded with people testing things and so much is flooded with attempt of certain scene building or motifs where one-on-one conversations are happening, I felt like I had the aha moment when I actually recognized that the character development as I'm implementing myself speaking or myself singing being a lover of poetry and of singing and of all different forms of art. I recognize the power and the strength of it being able to translate through the character that I was using within Hedra. I called it the soul behind the eyes. Wow, I really felt the uncanny valley fade away here. Even though I knew it was just an animated character, I totally believed them. That's kind of when I discovered, I don't at any point find myself outside of using Hedra, I always return. We're always returning to Hedra.

What are some types of projects you use Hedra for?

As some may have seen, and we can kind of highlight this one a lot, is that we partnered on one of our narrative shorts called the Original. And this one is showcasing the effect that we are world builders, that we do indeed care about the systematic process of where natural filmmaking meets AI - and we want to be the folks that bridge that gap. We're very proud of the Original series, as we call it, as we continue to build on that.

Another one I would like to talk about is the work we have done for a global eyewear company that hit a wall in production, meaning they needed something fast. They had a 15-day window that naturally would have taken at least four or five months. And we were able to execute it and the clients were thrilled. We always go back to that one as our proof of concept piece for a campaign, and that one we were very, very proud of because, again, these were global clients and made lots of cool new friends in Amsterdam.

Are there any other campaigns you have used Hedra for specifically for marketing purposes?

Certainly. I can even get more granular on the types. Some examples are social first campaign videos, character-led promotional content, meaning clients who want to use a mascot. That seems to be a theme as of late. Think using a Gecko like in the Geico campaign. We've had lots of clients that see character-driven content as a need. Other types are teasers, trailers, proof of concept videos, lots of those come down the pipeline for all different e-com clients.
Short form vertical clips. That's always part of the bread and butter. Investor-facing or partner-facing visual materials. That has definitely been the front runner as of late.

Do you find that animated characters are more popular than ‘real characters’?

I like this question because I feel it depends on the industry. And I'm not dodging the question. I've thought about this myself. I feel like the animation side be it more anime or be it an animation of a little character; they're more successful as the mascot for e-com products. At this point, in my, not so humble opinion, we're at, maybe the 90% of it being hyper-real perfection.
We've come very close, if not almost there. But the idea of animation as a mascot for brands, that's always been kind of cute and clever and fun. But it's also safer until the materials are understood, the legality behind the realism is understood. Having gone to a couple different conferences with the team, we also pride ourselves on paying attention to the weather of how to make sure that we educate ourselves and educate our partners on how to be secure as we use 3D models that we build. Let's say you are building a character from scratch, but yet, you want to make sure that it doesn't resemble an actress or an actor out there. We're very cognizant of that.

How has Hedra changed the way your team produces content?

When it comes to something I could translate to other agencies that are either interested in entering this market or want to try things out, I’d say the first milestone is to understand the value of the internal iteration process. If you have a grandiose concept, and you can explain it verbally that's great. But what's even better is to show a small clip or an emotional motif with some Suno music. That's the absolute winning ticket. Before you put a big external budget on a project, you can discuss and decipher how it actually could look like in real time. You have enough proof of concept without breaking the bank to show to your client. There's more of a vision. The truth is not everyone has that kind of imaginative thought process. They have to see it and that's normal. I think that is a great value of Hedra.

When pitching to clients, do you find that using AI for visual concepts is actually more valuable than using it for the final video?

It really depends on the industry and client. We're working with a fairly famous confection company out here in Nashville. They're the top seller on Amazon within their field. Ever since we've come on board, we've streamlined the iteration process giving them the flexibility of testing different variations, which is rad. But even more so because of that we've actually been able to attract a different demographic for the candy. Because we can iterate so quickly, it expanded their numbers, expanded their net. Now, they're in multiple, if not all the different grocery stores. It's kind of cool to see.

Was there a particular content that led to really strong business results for example new clients or high engagement?

We have a nice dichotomy of us showcasing these bite-sized pieces that are brand focused or narrative where it has naturally expanded our network of artists that want to work with us. But I was going to refer to the confection company for the specific question only because those are real metrics. Those are real numbers. At present, what's very exciting for us, is that we've done a significant amount of RFPs with different companies, for example entertainment companies helping them with pitching the 3D notion of what the events space will look like. We help with a lot of managers meetings, 50th anniversary events for big names in Chicago and in LA. We get to go in and do a quick rendition. It saves them a lot of time, is budget friendly and works as a quick iteration.

How do you use AI tools in the RFP process?

We just executed a nice 52-page brand deck that showed frame by frame how the person attending the event would feel from the first touchpoint of meeting the community that's greeting them all the way to the game between the events and then the end.

We were given blueprints that were pretty rudimentary. We infused them with some of the brand colors and themes, and we were able to present an elevated visual of what it could feel when you're there. We also rendered little clips where we put music, the brand sound, behind it. So they received a beautiful deck in tandem with a visual that showcased how it would actually feel to attend the event. That was very cool.

What do you find yourself going back to Hedra for mostly?

I think the biggest selling point from the beginning for me is speed.

When it comes to modern AI video tools, my natural workflow leans heavily toward platforms like Seedance 2.0, Kling, and Sora. The value here is all about the speed of iteration. These tools allow us to essentially 'throw paint on a canvas.' It’s a sandbox where we can push creative boundaries, test different figurative and literal 'lenses,' and see what sticks before we hone in and get meticulous with our own style. A lot of creators get frustrated with AI because they expect the perfect output on the first try. But the real secret sauce, and something we’ve experienced with platforms like Hedra and Higgsfield is that you have to build a relationship with the tool. You are building a conversation and a shared history within that specific generation thread. That history is exactly why we keep returning to Hedra.

When an AI video generation doesn’t hit the mark on the first try, what’s your process for fixing it? Do you have any secrets for tweaking the output?

I nodded to this a little bit but I feel patience is definitely of the essence when you're testing different things. I discovered that less is more; less is more in the written word. It doesn't mean, you let it hallucinate and decide for you. It's more getting as concise as possible on what you're going for conceptually. The best results come when I speak to it as if I'm directing an agent, you're essentially directing the director.

I've spoken to a lot of creatives about this, and what’s really interesting is that you don't need to feed the AI William Blake poetry to get something unique. But if you speak to it with clear intentionality, using phrasing that is a bit more abstract, surreal, or bizarre, it opens up this great middle ground. It lets you explore that creative space while still being incredibly strict about your camera types and lenses. I’m also a huge fan of prompting for multi-shots first. That allows us to iterate on the specific framing that’s working, and then we can take that exact same verbiage and lock it down into a single shot. It might not be a groundbreaking concept, but it’s definitely a core part of our secret sauce for tinkering and pushing these tools.

Where do you see AI video going for marketing teams in the next five years?

For those new to the space and for those that have been in the space, we heard it so many times, it's just getting better and better, and it's hard to keep up with all of those things. What I will say is that your process is the most important of them all.

People always talk about the big-picture stuff for marketing teams, things like eliminating creative restrictions and unlocking a team's potential to hit their goals. And sure, that's great. But realistically? The real magic is that it creates this perfect storm for whoever is leading the creative. It’s where your personal taste, your team’s taste, and your client's actual needs all meet in the middle. As for where the technology is going? Honestly, we can't even look at it quarter by quarter or month by month anymore. In my opinion, this space is changing week by week.

All those different marketing materials and pipelines we mapped out earlier are always going to be in flux. But I think this hybrid approach doesn't just apply to things like narrative filmmaking or acting, it applies directly to marketing and brands, too. My advice is: don't lose your foundation. Look, we have plenty of artists on our team who might sing a different song, but for me? After Effects is my bread and butter. No matter how advanced these tools get, I think I’ll always go back to it to lock things in.

AI ultimately is an assist. It's a tool. It's here to help you. It's not a full replacement.

Are there any potential limitations that you see with AI video as an industry?

I feel like it's not a matter of if, but when. For me, this always ties back to my obsession with building out a scene you can truly feel, where you actually feel the emotional interaction between characters. Whenever I consult with my mentors and industry peers on the pieces we're making, we always have these exact same conversations. Some of them look at it and go, 'Wow, this is fantastic, it’s 95% of the way there.' But others who have been in the industry for a long time will say, 'Yeah, but it could still be better here or there.'

To test those limits, we recently did a romantic piece where we purposely leaned into the elements that have always been notoriously hard for AI: hands and mirrors. We started the shot on their hands, building this incredible emotional and romantic tension. But when it came to the mirrors, working inside Hedra, I really noticed that it was missing those last 10% of realism. For instance, in the main shot her dress looked completely normal, but in the reflection, the AI rendered it as if the back of her dress was totally wide open. I was like, 'Wait, that’s not right.'

So, are there potential limitations? Oh, absolutely. But I’ll tell you, it changes so fast week by week that my main focus right now is just keeping up with it all.