AFK.live

Culture / Entertainment

Contact & Location

Phone

Email

AFK.live

General Overview

AFK.live is a French digital platform specializing in the aggregation, analysis, and automated indexing of news from over 250 French and international news sources. Its mission is to provide its users with "the signal without the noise" by offering an editorialized press review, structured by themes and clusters, allowing for a cross-sectional and pluralistic reading of information. AFK.live positions itself as a tool for monitoring, understanding, and discovering current events, avoiding algorithmic biases and the filter bubbles typical of social networks. The main site is accessible at https://afk.live/fr.

The company is published by kairos.io (SASU, SIREN 992 408 443), founded and directed by Sylvain Artois, a machine learning engineer and freelance CTO. AFK.live is the result of a personal project, developed outside of working hours, with a strong conviction about the necessity of supporting field journalism and combating misinformation (source: About page https://afk.live/fr/about-us/).

History and Origins

  • Launch: AFK.live was launched in October 2024 (source: About page).
  • Motivation: The project was born from a dual observation:
    • Field journalists are essential for establishing facts and combating misinformation, in contrast to often criticized editorialists.
    • Social media algorithms trap readers in informational bubbles, exacerbating polarization and ignorance of the other side.
  • Development: AFK.live was designed and developed by Sylvain Artois through his company kairos.ing, using machine learning tools and methods applied to information (source: About page).
  • Evolution:
    • Increase from over 200 to 254 indexed sources daily between 2024 and 2026.
    • Paris edition, version 2.0 of the site in May 2026 (major redesign).
    • Deployment of new editorial features (clusters, threads, files, semantic analyses).

Products and Services

AFK.live Platform

  • Main website: https://afk.live/fr
  • Main function: Automated aggregation, indexing, and curation of news.
  • Coverage: Over 250 French and international news sources, continuously updated.
  • Indexed articles: Approximately 5,000 to 5,500 articles per day (source: homepage, clusters page).
  • Edition: France, with international expansion.

Main Features

Clusters
  • Definition: Automated groupings of articles addressing the same news topic, generated via the BERTopic model.
  • Objective: To allow direct comparison of media coverage of the same event by different newsrooms.
  • Data:
    • Number of active clusters: between 23 and 25 depending on the runs.
    • Number of articles per cluster: variable, up to 20 titles per topic.
    • Number of sources covered per cluster: up to 17 newspapers for the same topic (source: clusters page).
  • Examples:
    • Adoption of major laws in France.
    • International geopolitical events.
    • Environmental or economic crises.
Threads
  • Definition: Editorialized tracking of current topics chosen by the editorial team, with thematic framing and automatic nightly refresh.
  • Usage: Allows tracking the evolution of a topic over several days or weeks (e.g., Avignon Festival, Tour de France, War in Ukraine, Forest Fires, Presidential Election 2027).
  • Number: Approximately 12 active threads, including 7 pinned (long-running) and 5 ephemeral (source: threads page).
Topics
  • Definition: Cross-thematic groupings, regularly updated, allowing navigation by major issues (politics, environment, tech, society, etc.).
  • Objective: To offer automated and structured curation of news by major domains.
Files
  • Definition: In-depth and multimedia analyses on complex issues, constructed from varied sources (articles, podcasts, books).
  • Examples:
    • "Europe in the LNG Trap: From Hormuz to the American Trap" (geopolitical analysis of European energy dependence).
    • "Dark Enlightenment vs. Green Lights: The Ideological Face-off" (mapping of neoreactionary and ecological currents).
  • Methodology: Files structured in thematic nodes, with excerpts, think tank analyses, podcasts, reference works (source: files page).
The Weight of Words
  • Definition: Chronicles of semantic and rhetorical analysis of public speeches (speeches, interviews, opinion pieces).
  • Methodology: Use of seven NLP indicators (entities, themes, Latourometer, readability, moral register, emotions, arguments).
  • Objective: To dissect the argumentative mechanics and moral register of key political and media actors.
  • Number: 14 chronicles published, covering 11 personalities and 3 source newspapers (source: The Weight of Words page).
Advanced Search
  • Functionality: Semantic search engine allowing users to find topics, keywords, or entities across all indexed sources.
  • Targeting: Search by topic, by cluster, by thread, by topic.
Methodology and Transparency
  • Editorial Pipeline: Automated, based on open-weight models (Mistral Large / Codestral) for entity extraction, framing classification, and synthesis writing.
  • Sources: Complete and transparent list of indexed sources (https://afk.live/fr/sources/).
  • Citation Limitation: Capping at 40% of the original text per citation, with systematic referral to the original source to encourage subscriptions and funding for field journalism (source: About page).

Business Model

  • Pricing Positioning: Premium offering.
  • Access: The platform offers part of its services for free, with advanced features reserved for subscribers (pricing data not public as of September 8, 2026).

Value Proposition

  • Signal without the noise: AFK.live aims to extract and present essential information, avoiding redundancy, paraphrasing, and editorial pollution.
  • Editorial Pluralism: Voluntary indexing of very diverse sources, from anti-capitalist left to republican right, to ensure a complete and nuanced view of current events.
  • Anti-Algorithmic Bubble: Refusal of opaque recommendation logics of social networks. Users are exposed to contradictory viewpoints.
  • Support for Field Journalism: AFK.live does not publish original news content but aggregates and systematically refers to media that fund investigation and reporting.
  • Automated and Editorialized Curation: Combination of ML pipelines and human intervention to ensure relevance, diversity, and structuring of content.
  • Methodological Transparency: Detailed explanation of models, sources, selection criteria, and limits of aggregation (source: About page).

Target Audience

AFK.live primarily targets two main types of profiles, formalized into personas:

  • The Informed Watcher

    • Information professionals (journalists, analysts, communicators).
    • Need to follow current events in a structured and synthetic manner.
    • Seeks time savings, quick identification of emerging trends from reliable sources.
    • Overwhelmed by the volume of information, values automation and reliability of curation.
  • The Curious Citizen

    • Public interested in current events, without the time or desire to browse dozens of sites.
    • Seeks a clear and neutral summary of important topics, free from editorial noise or opaque algorithmic recommendations.
    • Values transparency about sources and respect for professional journalism.

Typical Use Cases

  • Quick reading of the daily news summary.
  • Consultation of a synthetic dashboard by theme (dashboard, not infinite feed).
  • Tracking a complex topic via an editorialized thread or file.
  • Searching for a plurality of viewpoints on the same event.
  • Analyzing political or media discourse through the "The Weight of Words" chronicles.

Market and Positioning

  • Sector: Media, strategic monitoring, news aggregation, editorial intelligence.
  • Geographical Area: France (main edition), international opening via foreign sources.
  • Competitive Positioning:
    • No direct competitor mentioned to date.
    • Distinguishes itself from classic aggregators (Google News, Feedly) by its editorialized dimension, methodological transparency, and refusal of paraphrasing.
    • Offers an alternative to social networks and automated curation platforms by providing a plurality of sources and rejecting algorithmic confinement.
  • Target: Information professionals, engaged citizens, institutions, companies in strategic monitoring.

Team and Governance

  • Founder and Leader: Sylvain Artois, ML engineer, developer, freelance CTO (source: About page).
  • Legal Structure: SASU kairos.io, capital of €2,000, SIREN 992 408 443.
  • Legal Address: 61 rue de Lyon, 75012 Paris.
  • Hosting: Scaleway, 8 rue de la Ville l'Évêque, 75008 Paris.
  • Contact: contact@afk.live / sylvain.artois@afk.live / +33000000000.

Technologies and Methodology

  • Processing Models:
    • Mistral AI (open-weight models Mistral Large / Codestral) for entity extraction, thematic classification, NLP synthesis.
    • BERTopic for semantic aggregation of clusters.
  • Automated Aggregation:
    • Nightly re-indexing pipeline of sources and refreshing threads and clusters.
    • Use of the Brave Search API to broaden coverage beyond direct sources.
  • Transparency and Ethics:
    • No third-party trackers, no advertising cookies, no Google Analytics.
    • Minimal storage of personal data, only for technical functioning.
    • Respect for the right of reply for indexed press titles (source: About page).

Commitments and Values

  • Support for Professional Journalism: Refusal to aggregate content from social networks or non-journalistic sources.
  • Pluralism and Diversity: Indexing of sources covering the entire political and ideological spectrum.
  • Editorial Neutrality: Writing in the third person, refusal of partisan editorialization.
  • Transparency: Methodology, sources, models, and criteria explicitly detailed.
  • Respect for Copyright: Limited citations, systematic referral to the original source, encouragement to subscribe to media outlets.

News and Notable Projects

  • Continuous Evolution of the Number of Sources: Increase from 200 to 254 indexed sources between 2024 and 2026.
  • Deployment of Multimedia Files: In-depth analyses on major geopolitical, ideological, or scientific topics.
  • Chronicles "The Weight of Words": Semantic analysis of political speech, a unique tool in the French media landscape.
  • Major Redesign of the Site in May 2026: Improvement of ergonomics, editorial structuring, and transparency (source: About page).

FAQ – Frequently Asked Questions

Q: What is AFK.live?
AFK.live is a news aggregation and automated curation platform, indexing over 250 French and international news sources to offer a structured and pluralistic press review.

Q: Who is behind AFK.live?
The project is designed and developed by Sylvain Artois, a machine learning engineer and freelance CTO, through the company kairos.io.

Q: How does AFK.live select its sources?
Sources are chosen for their journalistic quality and ideological diversity, covering the entire political spectrum, from anti-capitalist left to republican right.

Q: Does AFK.live produce original content?
No, AFK.live does not write news articles. The platform aggregates, indexes, and structures the work of professional newsrooms, systematically referring to the original source.

Q: What is the difference between a cluster, a thread, and a topic?

  • A cluster groups articles on the same event or news topic.
  • A thread follows the evolution of an editorialized topic over time.
  • A topic groups articles by major cross-thematic issues.

Q: How does AFK.live combat filter bubbles?
By voluntarily indexing very diverse sources and refusing any opaque algorithmic recommendation logic, AFK.live exposes its users to contradictory viewpoints.

Q: Is AFK.live free?
Part of the services is accessible for free. Some advanced features and premium curation are reserved for subscribers (pricing not public as of September 8, 2026).

Q: How does AFK.live protect personal data?
AFK.live does not store any browsing data beyond what is technically necessary, does not use third-party trackers or advertising cookies, and user accounts are opt-in (source: About page).

Q: How can I contact AFK.live?
By email at contact@afk.live or sylvain.artois@afk.live, or via the contact form on the site.

Q: Can I request the removal or modification of an indexed article?
Yes, any press title can request the removal or modification of an article via contact@afk.live. Response within 48 hours on weekdays.

Q: Does AFK.live cover local news?
The platform primarily covers national and international news, but some local sources are indexed based on their relevance.

Q: How are political speeches analyzed?
Through the "The Weight of Words" chronicle, which uses NLP indicators to dissect the moral register, readability, arguments, and emotions of speeches and interviews.

Q: What are AFK.live's ethical commitments?
Respect for copyright, support for field journalism, methodological transparency, editorial pluralism, and protection of personal data.

Q: Does AFK.live offer an API or tools for professionals?
Not publicly mentioned as of September 8, 2026.

Q: Does AFK.live publish a newsletter?
Not publicly mentioned as of September 8, 2026.

Q: Where can I find the complete list of indexed sources?
The list is available on the page https://afk.live/fr/sources/.

Legal Mentions and Contacts

  • Trade Name: AFK.live
  • Legal Name: AFK.live / kairos.io (SASU)
  • SIREN: 992 408 443
  • Address: 61 rue de Lyon, 75012 Paris, France
  • Email: sylvain.artois@afk.live / contact@afk.live
  • Phone: +33000000000
  • Website: https://afk.live/fr
  • Social Media:
    • Instagram: instagram.com/afk.live
    • LinkedIn: linkedin.com/company/afk-live
    • Bluesky: bsky.app/profile/sylvain.artois.me

Summary

AFK.live establishes itself in 2026 as a reference platform for monitoring and understanding current events, combining algorithmic power, editorial pluralism, and respect for professional journalism. Its transparent methodology, refusal of algorithmic bubbles, and commitment to diversity of viewpoints make it a valuable tool for information professionals as well as for discerning citizens. Its entire offering, structured around clusters, threads, files, and semantic analyses, meets a growing demand for synthesis, structuring, and reliability in a saturated and polarized media landscape.

Main Sources:

This listing is structured according to the Wispra methodology — Our approach →