• by Laurie Sullivan , Staff Writer @lauriesullivan, December 30, 2024

Akkio, an AI data platform for advertising agencies, earlier this year released what its CEO calls the first AI data platform and large language model (LLM) for media agencies.

It enables agency reps -- like the ones at Horizon Media -- to talk with campaign data in natural language to build audiences, create visualizations, and uncover insights. The data comes from the brand, websites, credit card transactions, and more.

“It takes hundreds of billions of rows of data across tens of thousands of columns of information, but you can operate across all with natural language,” says Akkio CEO Jon Reilly. 

“Reporting on the performance of a campaign at the end can take up to four weeks,” he said. “With AI tooling and creation integrated directly into data flows, it can take an hour or less.”

The data engine gathers all the context and plugs in the best-performing model, explains Reilly, who began building a LLM engine and operations modules to classify and forecast need so leads were scored as they came into the company.

At some point the challenge became processing the data and then bringing it back into platforms that its developers began building. Multimodal images now are available when building an audience.

Akkio works with about 15 agencies. The platform is open, so several other marketing teams also use it.

“It takes hundreds of billions of rows of data across tens of thousands of columns of information to build audiences, but you can operate across all with natural language,” he says.

The other day when playing around with the system, Reilly told the platform to build an audience of “gun-toting ranchers in New Mexico who vote Blue.”

It seemed like a crazy request, he said, but the system did it because it had all the demographic datasets to accomplish the task.

Reilly describes himself as a "product guy," but he has experience working at two major audio-focused companies, Sony and Sonos. “Manual tasks created a terrible experience,” he said. “We started thinking about how to do these tasks within the platform.”

Before OpenAI released ChatGPT, Reilly said company developers got the idea to use generative AI to write the code.

Akkio built the platforms Chat Data and Chat Explore, and others followed. Giant data engines today sit on top of Snowflake that communicate datasets, with hundreds of billions of rows of data that allow the average user to just ask for what they want.

The company builds workflows that can automate and replicate manual tasks. For example, a user can ask the platform Agency Data LLM to build a specific audience, and the technology will go out into Akkio’s datasets to find the data and build it.

It also will tell the advertiser everything they need to know about the data that sits in the audience segment. Mapping technology divorces the identity of the individual and looks for patterns to target audiences, he says, to maintain privacy.

Reilly says the company in 2025 will work on automating all different types of processes, including audience planning and exploration, to campaign adjustments and real-time reporting that identifies cost per conversion exceed a dollar without writing an "alert" line. "Soon you will be able to ask the platform in natural language, 'tell me when my cost per conversion exceeds $1."