Home AI This Company Says It’s Found A Way To Balance Automation With Human Control

This Company Says It’s Found A Way To Balance Automation With Human Control

SHARE:

Brands are being constantly barraged with the message that agentic AI is the future of advertising.

But marketers are struggling with the question of how much power to put in the hands of a machine.

Making Science believes it’s hit the sweet spot.

The digital marketing consulting firm recently launched a tool called Creative Hub that helps advertisers optimize their marketing campaigns by bridging the gap between creative and execution while also providing brands with full control over their creative versioning.

Creative Hub is part of ad-machina, Making Science’s agentic AI platform.

Making Science has seen some resistance to generative AI because marketers feel a lack of control over brand guidelines, according to Costanza Ghelfi, chief product officer at Making Science.

Generally, brands have a performance team and a creative team. What’s often missing, Ghelfi said, is technology that “speaks the language of both.”

The best (re)version of yourself

Creative Hub allows brands to analyze current ads to determine what is and isn’t working and revise accordingly.

For example, Ghelfi said, one client saw that ads set in daytime performed noticeably worse than ads in nighttime settings. That brand worked with Making Science to update its preexisting ads so they appeared to be happening at night, while ensuring that nothing else about them was removed or revised.

But marketers can also use the tool to make more significant changes to ad creative, including adding net new elements, like animals or flowers or anything, really, so long as the ads perform better with those ingredients.

Subscribe

AdExchanger Daily

Get our editors’ roundup delivered to your inbox every weekday.

Creative Hub mandates a human’s approval before making any updates to an ad, and that isn’t going to change, even as the technology evolves, Ghelfi said. Advertiser control is “core in the value proposition,” she added.

Although Creative Hub can generate new visual elements and reversion existing campaigns, it “doesn’t generate anything from scratch,” said Ghelfi, and “it’s not in our road map” to include that as an option.

Plus, sticking to material that is already client-approved is part of how Making Science stays compliant with the EU AI Act – a “very restrictive” act, according to Ghelfi, of which the European-based company must stay abreast.

Lightening the load

Still, any use of generative AI in creative sparks concerns about how it will affect human roles.

Ghelfi insists that Making Science has only seen positive results on both the human side and the product side. “It’s allowing us to be much more efficient, to innovate much faster,” she said.

This has been the refrain of every company in the past year – including those who have undergone mass layoffs, like Business Insider and Cisco.

But Ghelfi emphasized that Making Science views agentic AI as a tool for growth, rather than for cost-cutting. The latter, she says, is likely just a “short-term effect” of early AI adoption.

The companies that will thrive in the long run are those that see AI as a way to innovate and expand their teams, she said, rather than to cut corners. “We are hiring much more now,” she added, “because we are able to create more, to sell more.”

When asked how Making Science is using that extra time – Creative Hub has doubled the company’s productivity, according to Ghelfi – her response was that, well, they don’t.

“We’re not offering a specific alternative to what they can do with that time,” she said. “They live better. They leave the office earlier.”

Oh, European sensibilities.

Ghelfi acknowledges that the proliferation of AI has “been disruptive” to many companies, but now that there are end-to-end AI marketing solutions and less expensive ways of developing new tools, she’s hopeful that more companies will approach AI the way Making Science does. As in, to help ease workloads while expediting production and innovation.

“I’m one of the optimistic ones,” she said.

Must Read

Amazon Ads Is All In On Simplicity

“We just constantly hear how complex it is right now,” Kelly MacLean, Amazon Ads VP of engineering, science and product, tells AdExchanger. “So that’s really where we we’ve anchored a lot on hearing their feedback, [and] figuring out how we can drive even more simplicity.”

Betrayal, business, deal, greeting, competition concept. Lie deception and corporate dishonesty illustration. Businessmen leaders entrepreneurs making agreement holding concealing knives behind backs.

How PubMatic Countered A Big DSP’s Spending Dip In Q3 (And Our Theory On Who It Was)

In July, PubMatic saw a temporary drop in ad spend from a “large” unnamed DSP partner, which contributed to Q3 revenue of $68 million, a 5% YOY decline.

Paramount Skydance Merged Its Business – Now It’s Ready To Merge Its Tech Stack

Paramount Skydance, which officially turns 100 days old this week, released its first post-merger quarterly earnings report on Monday.

Privacy! Commerce! Connected TV! Read all about it. Subscribe to AdExchanger Newsletters
The Arena Group's Stephanie Mazzamaro (left) chats with ad tech consultant Addy Atienza at AdMonsters' Sell Side Summit Austin.

For Publishers, AI Gives Monetizable Data Insight But Takes Away Traffic

Traffic-starved publishers are hopeful that their long-undervalued audience data will fuel advertising’s automated future – if only they can finally wrest control of the industry narrative away from ad tech middlemen.

Q3: The Trade Desk Delivers On Financials, But Is Its Vision Fact Or Fantasy?

The Trade Desk posted solid Q3 results on Thursday, with $739 million in revenue, up 18% year over year. But the main narrative for TTD this year is less about the numbers and more about optics and competitive dynamics.

Comic: He Sees You When You're Streaming

IP Address Match Rates Are a Joke – And It’s No Laughing Matter

According to a new report, IP-to-email matches are accurate just 16% of the time on average, while IP-to-postal matches are accurate only 13% of the time. (Oof.)