The New York Times is pleased with the performance of its GenAI ad-targeting experiment BrandMatch. After running a beta test with six brands across tech, finance, luxury, and other categories, the news publication is now inviting other advertisers to try BrandMatch as well.

“BrandMatch works by building personalised targeting segments for every ad campaign, capturing the nuances of a brand and its audience,” a Times spokesperson said. “With this new tool, advertisers can provide their existing marketing briefs, which BrandMatch uses to build personalised ad targeting segments based on relevant articles and the audiences that most engage with those articles.

“BrandMatch provides a solution to a common challenge that marketers face: how to reach specific target audiences as described in their briefs when traditional targeting can only offer a predefined menu of targeting criteria to choose from. The new tool works in a different way than existing ad targeting models, automatically interpreting a brief and matching it with the most relevant context and audiences.”

Meanwhile in Denmark, news magazine Dagbladet’s news magazine Information has built a chat product on top of a new customer database, which has been 10 months in the making.  

“Now I and others in our organisation can chat directly with the underlying tables of data in pure prose,” said Chief Commercial Officer Simon Fancony. “This will save me immense amounts of time and — more importantly — save a lot of time for our already extremely busy data analysts who have spent so much time writing SQL-queries and making dashboards for us non-SQL/data-savvy people to better understand our business in the past. Hopefully, we have now reduced that time significantly.”

The Washington Post has built an AI tool called Haystacker that analysed more than 700 political ads to help with an investigation. 

Haystacker uses the vision capabilities of a large language model. It extracts stills from video files, processes them to on-screen text, and labels the objects present, such as an American flag. The text and visual information are then analysed and verified by reporters. The tool, which took more than a year to build, can be used on any large dataset to find patterns.

And in New Zealand, Stuff Group has created a customised AI tool that helps with the reporting of public documents such as council meeting minutes, submissions, and government reports. Journalists use the Democracy AI tool to scan, prioritise, and report on hyper-local decision-making documents. They still cross-check the content with source documents. 

“Our trials of this GPT tool, which uses official council and government documents as source data to scan for news and decision making, has shown how we can streamline some aspects of the news gathering and writing process, allowing our reporters to focus on polishing and adding depth to the stories,” said Stuff Masthead Publishing Managing Director Joanna Norris. 

The tool can edit stories to varying lengths and formats, and write summaries and headlines, for both print and digital.

At one Stuff publication, The Waikato Times, Editor Jonathan MacKenzie said he was amazed by the speed at which his reporters could work when assisted by AI. 

“We have 11 local authorities in our coverage area so this tool is a win-win for our widely dispersed audience and the newsroom,” he said. “It’s far better for my staff to be out talking to people and digging for stories than stuck behind a desk reading a council agenda.”

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .