Skip to content
AI Radar by Wagner Solution
CS
Menu

About the project

One clear overview from dozens of sources

My name is Patrik Wagner and I have been actively following AI for a long time, trying new tools and looking for practical uses for them. But the number of sources keeps growing, and staying truly informed is becoming a full-time job.

That is why I created AI Radar by Wagner Solution: one place that tracks important developments in AI, compares available information and prepares a clear overview in English.

72 active sources 436 published events

Open today’s briefing → How AI Radar works ↓ Message to readers: AI Radar refreshed: a more readable website and further quality improvements →

Why it was created

Less information noise, more context

Problem

One change, many similar articles

Alongside important news, there are reprints, marketing announcements and content with no practical impact. What matters can easily get lost among the many sources.

Solutions

One event with a summary and original sources

When several websites cover the same thing, AI Radar brings their coverage together, explains the significance of the change and preserves links to the originals. The result is not an automated reprint or another RSS reader.

From signal to event

From source to a useful overview

First layer AI Flash

Brief news items link to the original; original analyses identify AI Radar as the author.

 Second layer Today’s briefing

A verified event with context and practical impact.

For price changes, after checking the official price lists, we prepare a separate article calculating the impact. The original news item retains the manufacturer's facts; our own interpretation, charts and methodology are clearly separated.

  1. 01
    Monitors

    Checks curated sources, official announcements, research, regulation and security.

  2. 02
    Selects

    Filters out content with no real connection to AI, marketing noise and old reprints.

  3. 03
    Connects

    Combines articles about the same thing into a single event. Ambiguous cases are not merged automatically.

  4. 04
    Summarises

    Assesses the significance of the change and prepares a brief summary, translated into English, with links to the originals.

  5. 05
    Explains

    Turns the same facts into practical implications for individuals and business decisions.

AI Radar under the hood

Works even when you are not reading

It continuously reviews sources on its own VPS, groups articles about the same change and provides context in Czech and English.

How the system works →

Today's AI work

9.59M

tokens today

Exactly 9,594,138 · 954 completed runs

Total input, output and cache usage reported by AI models since midnight in Prague.

Source sorting What is relevant and worth further coverage
559,311 28×
Linking events Multiple articles about the same change grouped into one topic
419,397 41×
Summary and translation Czech context, impact and a practical next step
55,858 3×
Other processing Further automated steps in the editorial process
8,559,572 882×

Running services

Running on your own VPS

System running
01 HTTPS gatewaySecure website delivery online
02 Web applicationOverviews and public pages online
03 Worker and schedulerCollection, queue and AI processing running
04 Data layerPostgreSQL and vector search online

4/4 layers online Latest AI work just now

What you will find here

Everyone looks for something different in AI

Choose a view based on what you need to resolve right now.

Daily briefing

Today’s briefing

The most important changes without duplicates or information noise.

Open overview →

For myself

For individuals

New tools and features you can use in your work.

Find a practical use →

For decision-making

For businesses

Impact on processes, costs, risks and options for adopting AI.

Explore business impacts →

For specific work

Model selection

Comparison by use case, capabilities, price and speed.

Choose a suitable model →

Find ongoing developments in AI Flash. You can continue browsing completed events by topic, in the archive or separately in research.

How we handle confidence

The more important the claim, the stricter the verification.

We publish significant news events only after independent confirmation. For an unambiguous product release, legislation or operational incident, the relevant official source is sufficient. A narrowly focused technical guide may rely on a single source, and its card will always label it as such.

Current publication review status

44 of 44 significant events meet publication rules

We count the actual publishing rule, not just the number of articles or feeds.

  • 2 independent sources

    At least two independent publishers have confirmed the claim. The card shows their number.

  • Official announcement

    A product release, legislation or an incident is substantiated by the relevant authority.

  • 1 source

    A regular news item or technical guide has not yet been independently confirmed. The card explicitly indicates this.

Current coverage of standard content

71 of 471 regular events and AI Flash items have corroboration from multiple sources or an official source. For the remaining 400 one source is visibly listed.

How we calculate evidence status

We consider an event independently confirmed when it is documented by two independent publishers, or by an official source and another independent publisher. A single official source is sufficient only for an unambiguous product release, legislation or operational incident.

Multiple feeds from the same owner count once. AI aggregators and automatically generated overviews are not used as evidence and are not listed among active public sources.

Quick overview

How to read an event card

The main selection uses cards; other lists use compact rows. Four elements help you find your way.

  1. 01
    Significance

    Our importance score from 0–100 expressed in words. In listings, it is shown with bars and a word: “major”, “important”, “monitoring”. The event card and detail page clarify the lower level as “worth attention” or “context”.

  2. 02
    Trust

    The line with ✓ states what supports the news item: an official source, independent confirmations, or just a single source so far. The card always makes this clear.

  3. 03
    Impact

    Practical impact appears in the detail and the For individuals and For businesses views. The main selection may show it directly on a card; a compact list row shows the shared summary.

  4. 04
    Sources

    “Details and sources” leads to the full summary and all original articles. Verify important decisions against the originals.

AI's role

AI helps process the volume. It is not a source of facts.

The model classifies articles, prepares a Czech summary and drafts an assessment of practical impact. It must not add facts from its own memory. The output is checked for empty or test content, language errors, inflected names and potential duplicates.

What the score means

Importance is not popularity.

We assess impact, usefulness, novelty, credibility and urgency. The number of distinct publishers gives the result a slight boost. We use verbal levels publicly because a number alone, without context, creates a false sense of precision.

Limitations

Any summary may be incomplete.

AI may miss context, and sources may be wrong or change later. That is why we list the original articles for each event and recommend checking the primary source for important decisions.

Under the hood

Your own server. Multiple models. One managed process.

AI Radar is not a single chatbot with a long prompt. An automated processing pipeline runs on my own server: at regular intervals, it checks curated sources, loads new articles, cleans the data and compares whether different websites are reporting on the same change.

A fast model first filters out noise and estimates the information value. Another model works only with what matters: it compares sources, prepares a Czech summary and describes the practical impact. Between steps, the system checks trustworthiness, duplicates, language and publication rules. AI is not a finishing touch here, but part of the infrastructure from the first signal to the finished overview.

Transparent inputs

Active sources and their roles

“Primary” does not mean independent. A manufacturer's announcement documents what it released; independent journalism, research and institutions provide a different kind of evidence. The list does not include discovery aggregators.

Community signal

5 sources

Primary source

25 sources

Operation and availability

11 sources

Independent editorial team

21 sources

Regulace

1 source

Research

9 sources

No jargon

Glossary

Canonical explanations of terms you will encounter on the website. We explain unfamiliar terms when they first appear — here they are all in one place.

Event
One thing that happened. We bring together all the articles covering it — so the website has one card, not ten similar news items.
Model
The “engine” inside an AI tool. The ChatGPT app is powered by a GPT model; the same model can run in different apps.
Source
A website we read articles from: an official provider announcement, an independent publication, a research site or an institution.
Importance
Our score from 0–100 for how much an event changes practice. In listings, we show it with bars and a word: “major” (80 or above), “important” (60–79) and “monitoring” (below 60). The event card and detail page specify the lower tier: “worth attention” (40–59), otherwise “context”.
Trust
What evidence supports the event — an official source, independent confirmations, or just a single source so far. The card always states this explicitly.
Breaking news
An unfolding event. The summary may become more precise as further sources become available.
Preprint (study before peer review)
Research published before other experts have had a chance to review it. We keep it separate from verified news.
Benchmark (comparative test)
A measurement of a model's capabilities on a set of tasks. It is a measurement, not a verdict — the result applies to that type of work.
Beta (trial version)
A feature released for trial use. It may change, have limitations or occasionally work unreliably.
Agent
AI that handles a task on its own across multiple steps: it reads, makes decisions and uses tools — rather than just answering a single question.
Open-source
A model or tool with publicly available code or weights. It can run on your own infrastructure.
Hallucinations
When AI makes up a plausible-sounding but false piece of information. That is why we link to original sources for every event.

For source operators

Considerate data collection and copyright

The project introduces itself with a header AIRadar/1.0 (+https://ai.patrikwagner.cz/o-projektu). It uses public feeds, conditional downloads and reasonable intervals. It does not log in, bypass paywalls or circumvent technical restrictions.

Full texts of articles by others are not published. They serve only as internal input for creating our own summaries; readers always continue to the original. If you would like a source removed, write to patrik.wagner@gmail.com.

The project as a living case study

What can you build with AI using your own data?

A similar approach can be used for market monitoring, internal knowledge, sales support, or decision-making based on larger volumes of information.

Wagner Solution ↗