Hello, this is Sergei.
This week there’s a lot to read. Probably more than usual. And while going through the newsfeed, LinkedIn, and BlueSky, I caught myself thinking about one shift that’s becoming obvious: journalists are using AI more — and in a more complex way.
A year ago it was mostly one tool. ChatGPT. Claude. NotebookLM. Now I see people building small systems. Combining models, creating layers of usage of the models. Connecting workflows. Some of those examples are in today’s digest of initiatives and resources.
Let me add my own five cents.
This week I rolled out a content analysis tool for one of my clients. It’s a small AI-driven system built around real performance data.
Here’s what it does.
It fetches content from Instagram, Facebook, Telegram, Twitter. It collects views and reach for every post, calculates performance scores, and applies weighted engagement metrics — likes, shares, clicks, comments. Then it selects the top 100 best-performing posts for each platform.
After that comes the AI part (I use combination of Groq's openai/gpt-oss-20b and Claude's claude-sonnet-4-5-20250929). The system reads the content of those top posts and identifies patterns — formats, structures, angles, recurring themes. It calculates average performance scores per pattern to show which ones clearly outperform others.
And finally, it generates practical recommendations: what to replicate, where to do it, and when.
Not theory. Not “AI will transform journalism.” Just applied analysis that helps a team make better decisions next week.
This is the direction I see more often now. Less experimentation with shiny tools. More building of small, focused systems that solve specific problems.
New on AI For Newsroom this week
Stories, guides, initiatives, and signals we surfaced in this issue.
Get the weekly email digest
One curated Friday roundup for newsrooms. No spam — you can unsubscribe anytime.