• by Laurie Sullivan , Staff Writer @lauriesullivan, Today

Artificial intelligence (AI) has a place in Ampersand’s business model, but today the company’s investment goes more toward machine learning for ad targeting that helps advertisers understand incremental reach and predictive models.

“The data we get from the 63 million households tell advertisers what was viewed,” said Mike Dean, president of Ampersand. “Not just the ads being viewed, but also the content.”

Dean expects Ampersand to use large language models (LLMs) and generative AI (GAI) to develop TV-ready creative that will give even the smallest advertisers a chance to use multiplatform video in streaming.

These smaller companies have not yet reached the TV market. Today they rely on social media to use video. 

“Our owners have already made investments to bring GAI to small businesses that do not have TV-quality creatives,” he said. “It’s getting buyers to think about other investments.”

Ampersand, a joint venture between Comcast, Charter, and Cox, represents data from about 118 million multiscreen households. It focuses on closing the gap between streaming, fragmented TV landscape, and inefficiencies in the TV buying process. The ad inventory comes from the cable networks.

To complicate matters, each joint partner -- along with Verizon Media, now known as Yahoo -- has their own infrastructure. Ampersand must combine all data to make the buying process fast, easy and seamless.

Combine the household data with audience data and the massive number puts Ampersand on par with a company like Roku.

“Most ad buyers have never heard of Ampersand,” Dean said, but the “data is sold into three independent ad buying markets such as political, multiregional advertisers, and national to support addressable like household level targeting.”

When asked about a space for agentic in Ampersand’s product roadmap, Dean provided a hypothetical example.

“We’re not planning for this, but a buyer could go to an AI interface like ChatGPT and type in their budget, brand and objects to get the best media plan to build,” he said. “Today that’s done manually, but the AI model would take in the information and spit out a plan.”

Automation of work that is done more efficiently by an algorithm could become the plan, but “we’re not investing in that space today.”

Dean said the data is more of a differentiator for Ampersand 00 rather than using GAI, which could determine the best media-buying, budget, and targeting options.

“In the pie chart of media, search is dominated by Google, social dominated by a handful of walled gardens, and you really have a multiplatform video ecosystem that’s incredibly fragmented,” he said. “That’s where Ampersand plays.”

Ampersand’s buyers expect high-quality data.

“If someone handed me $10 million to invest, I would probably invest it in data engineering, rather than building out AI capabilities,” Dean adds.