HOW IT WORKS

Copy the text for analysis (e.g., an article, a social media post) and paste it into the input field.

Select the primary language of the text (English or Ukrainian) using the buttons above the input field. This selection is required, as it determines which language-specific system prompt is used for the analysis.

Click the ANALYZE button. The request is sent to the secure Svitlogics API gateway. Your input text is never stored.

The Svitlogics back-end service utilizes a high-availability cascade of premium Google Gemini 2.5 models. This system ensures reliability: if a primary model is at capacity, the request automatically falls back to an alternative.

The selected model is guided by a detailed, custom-calibrated system prompt. The prompt instructs the AI to assess the text against five core criteria, functioning as an analyst, not a "truth detector":

• Manipulative Content
• Propagandistic Content
• Disinformation
• Unbiased Presentation (Impartiality)
• Emotional Tone

After a short delay, the system returns a structured report. The report contains a percentage-based score and a concise justification for each of the five criteria, along with an overall summary.

The results provided by Svitlogics are not a definitive verdict. Svitlogics is an auxiliary tool designed to aid your own critical thinking. Use the provided data points and analytical justifications to question the text's underlying messages, identify potential biases, and form your own, more informed conclusions.