Founded in 2012, naTemat is one of Poland's leading digital-native news publishers, reaching audiences through its news platform, YouTube channels and social media. In recent years, the company has made video a central part of its editorial strategy, investing in studio production, podcasts and platform-native formats. In this interview, Paweł Orlikowski explains why successful newsroom video is about more than publishing at scale, how shorts and long-form content complement each other, and why authenticity and audience engagement matter more than simply producing more videos
Let’s start with the basics. How is video production organized in your newsroom today?
We’re still in the growth phase when it comes to the technical team. Right now, five people handle production, editing, realization, and studio operations for our editorial content - by which I mean our own signature programs and podcasts.
At the same time, we’re scaling. We currently have two studios - one main studio and one more lifestyle-oriented setup, but the goal is to eventually operate three independent studios for parallel productions.
We also run several YouTube channels. Our strategy is twofold: grow the main channel while simultaneously building thematic channels around specific content categories.
What does the workflow itself look like?
Most productions are classic studio-based formats: recording, editing, distribution. The core platforms are YouTube and Spotify, though we’re expanding into additional partnerships and distribution channels.
A standard episode usually runs somewhere between 40 minutes and 1 hour 20 minutes. From every long-form conversation, we also create vertical short-form content, typically between 8 and 15 shorts per episode.
Today, virtually every platform favors 9:16 vertical video. Occasionally we create square formats for places like LinkedIn, but vertical dominates.
How much video do you publish?
Usually one or two long-form videos per day across our channels. That’s already a significant production effort for a relatively compact team. On top of that, we publish several to over a dozen shorts daily.
What matters to us isn’t simply volume. Publishing five videos a day sounds great, but if the audience can’t realistically consume that amount of content, or if it damages algorithmic performance, then it’s counterproductive.
Are you already using AI in production workflows?
Yes, but more as operational support than replacement - mainly for transcription, timecodes, and identifying potentially interesting fragments from longer conversations. Tools like GPT or Gemini help us work with transcripts and speed up editorial decisions.
But AI still struggles heavily with context, especially in Polish-language content.
The problem isn’t generating clips mechanically - many tools can technically do that. The problem is understanding what actually makes a fragment meaningful, emotional, or engaging.
We tested fully automated clipping workflows and honestly, many of them slowed us down more than they helped. Often the subtitles are burned in incorrectly, editing flexibility disappears, or the output simply feels off. Text-based assistance works better than fully automated editing.
If a perfect AI clipping tool existed, would you want it?
Absolutely. If I could upload a full program and get five intelligently selected clips with real audience potential, that would be incredibly valuable.
But we’re not there yet. Current tools still don’t truly understand narrative context or emotional weight.
You're 40% through.
The rest opens with an email
Free - there is no payment at any point. One email opens this piece and everything else in the report.
6 locked sections of this conversation
15 remaining interviews and commentaries
All four topic reading paths, end to end
The Sotrender data, and the complete PDF