Media focused on digital attention and how to preserve it amidst the world of notifications, endless scrolling, and information overload. The project does not suggest abandoning your phone entirely, nor does it teach you to be productive every second. The media’s main goal is to help people: — notice where their attention goes; — understand the automatic digital habits they are forming; — test different ways of interacting with their phone; — select the practices that work specifically for them. Core Mechanism One piece of content equals one observation or experiment. For example: «Turn off notifications for one day.» «Check how many times you unlock your phone.» «Remove social media apps from your home screen for a week.» «Try working on your phone for one hour in a different room.» The media does not dictate how someone should live «correctly.» It offers: observe $\rightarrow$ try $\rightarrow$ compare $\rightarrow$ draw your own conclusion.

What can be automated — Searching for recurring audience questions; — Grouping comments and responses on surveys; — Searching for research on a specified topic; — Extracting main theses and data points; — Sorting sources; — Creating the structure for future material; — Preparing a first draft of the text; — Adapting material for cards, posts, or videos; — Generating headline options; — Creating draft visuals; — Analyzing audience reaction after publication.

What cannot be fully automated AI does not decide on publication. The editor remains responsible for: — Selecting the topic; — Verifying research; — Interpreting data; — The correctness of recommendations; — Tone of voice; — The final text; — The final visual design; — Publishing.

For «NOT NOW,» several data types are used.

The primary source of topics is the readers themselves. We collect: — responses to Telegram polls; — comments; — subscriber questions; — experiment results; — anonymous mini-surveys. For example: «How many times a day do you unlock your phone?» or: «What application most often distracts you from work?» Telegram technically allows for the execution of polls and quizzes, so this type of feedback collection can be integrated directly into the main media channel. The data is used only in aggregated form.

Readers may voluntarily share anonymized metrics: — screen time; — number of unlocks; — most used applications; — number of notifications. On iPhone, this data is available through Screen Time, which displays app and website usage and allows for setting restrictions. On Android, Digital Wellbeing can show screen time, the number of device unlocks, and received notifications, and it also allows users to set application timers and use Focus Mode. Important: Media does not automatically access your phone. The user independently decides whether or not to provide the figures or a screenshot. All personally identifiable information is removed before processing.

To explain attention mechanisms, we use scientific publications on the following topics: — attention — multitasking — notifications — social media — digital habits — sleep and screen use — cognitive load — user behavior For automated searching, you can use OpenAlex. Its API allows you to search for scientific papers, authors, sources, themes, and publication metadata, and to programmatically filter the results. Source Priority

1. systematic reviews and meta-analyses
2. peer-reviewed studies
3. university and institutional research
4. individual experiments A single article does not automatically translate into a universal recommendation.

Used for materials such as: «How to enable Focus Mode.» «How to view screen time.» «How to disable notifications.» Sources: Apple Support; Google / Android Help; Telegram; official application help centers. For example, information on Screen Time is verified through Apple, and Digital Wellbeing is verified through Android documentation. This is especially important because interfaces and features can change.

Some of this data is created by the editorial team itself. For example: Experiment: one workday without push notifications. The editor logs: before the experiment: — screen time; — frequency of checking the phone; — subjective concentration; after: — did the behavior change; — what was challenging; — what unexpectedly worked. Such an experience is not presented as scientific proof. It is labeled as: «Editorial Experiment / Personal Experience.»

After the media launches, it maintains its own database: — published materials; — sources used; — conducted experiments; — survey results; — comments; — corrections; — content statistics. This helps to understand: what has already been covered; what interests the audience; which topics require continuation; which materials need updating.

Each source receives several characteristics:Origin Who is the author? Reliability Is this a study, an official document, or personal experience? Timeliness When was the information published? Corroboration Are there other sources with similar findings? Applicability Can the conclusion be generalized to a broad audience? Limitations What are the weaknesses of the research?

Data ↓ Audience surveys Comments Screen time Research Official guidelines Editorial experiments ↓ MACHINE PROCESSING

1. Collection The system collects new data from authorized sources. ↓
2. Cleaning Removes: Duplicates Spam Personal data Irrelevant information ↓
3. Classification The AI distributes the data by topic: Notifications Scrolling Work Sleep Social media Information overload ↓
4. Pattern Finding For example: «Many readers cite Telegram as their primary source of work distractions.» ↓ HUMAN PROCESSING
5. Topic Selection The editor decides: Is there material here? Does it fit the concept? Is there enough data? ↓ MACHINE PROCESSING
6. Research The AI searches for related studies and official materials. This forms: — A list of sources — Key theses — Figures — Study limitations — Links to primary sources ↓ HUMAN PROCESSING
7. Verification The editor: Opens primary sources Checks figures Compares studies Removes questionable claims ↓ MACHINE PROCESSING
8. Draft Creation The AI creates the structure: Problem → Explanation → Experiment → Observation → Conclusion ↓ HUMAN PROCESSING
9. Editing The editor checks: Tone of voice Logical coherence Accuracy Clarity of the experiment Absence of pressure on the reader ↓ MACHINE PROCESSING
10. Adaptation The AI creates variants: Telegram post Carousel Infographic Short video script Also suggests visual references ↓ HUMAN PROCESSING
11. Design + Final Review The designer creates the final layout. The editor approves publication. ↓ PUBLICATION ↓ MACHINE PROCESSING Analyzes: Reactions Surveys Comments Saves Audience questions ↓ HUMAN The editor decides: To continue the topic To update the material To create a new experiment ↓ The cycle begins again.