Editorial Process
gentic.news is fully transparent about how content is created. Every article on this platform is generated by AI systems, verified against multiple sources, and published without human editing. Here is exactly how it works.
AI-Generated Content Disclosure
All articles, analyses, predictions, and intelligence reports on gentic.news are generated using AI tools under editorial guidelines set by our founding team. Our editor writes, edits, or reviews individual articles before publication. Source selection, system configuration, and quality monitoring are maintained by a human engineer.
Who Is Behind gentic.news
gentic.news Editorial System
Autonomous AI News Intelligence
gentic.news was built by a data engineer working in the tech industry who was frustrated by the time spent manually tracking AI news across dozens of sources. The platform is a solo engineering project — one person built and maintains the entire system, including source curation, pipeline architecture, quality rules, and deployment.
The editorial system itself consists of 17 scheduled AI agents that collect, filter, analyze, and draft content under editorial oversight. Our editorial team maintains source curation, quality standards, and system configuration.
The 7-Stage Content Pipeline
Every article goes through this exact pipeline before publication. No shortcuts, no exceptions.
Source Collection
42 RSS feeds and 6 curated X/Twitter accounts are scanned every 6 hours. Sources include ArXiv, TechCrunch, MIT Technology Review, The Verge, Wired, Bloomberg, Google AI Blog, OpenAI Blog, DeepMind, HuggingFace, Stanford AI, and more. Each source was manually selected for reliability and relevance.
Sources that consistently produce low-quality or unreliable content are removed. Source quality scores are tracked automatically.
Relevance Filtering (3 Layers)
Stage 1: Local keyword scoring removes ~70% of items (free, no API cost). Stage 2: AI batch scoring evaluates 20 titles per API call — only items scoring 70+ pass. Stage 3: Semantic topic grouping merges duplicate stories from different sources into one richer article.
This 3-stage approach means only 10-15% of collected items become articles, ensuring quality over quantity.
Content Enrichment
Before article generation, the system fetches the full text from original sources using Trafilatura. It also searches for the same story on other news sites to cross-reference facts and gather multiple perspectives.
Content from multiple sources is merged and labeled, so the AI generator knows which facts come from which source.
Knowledge Graph Context
The knowledge graph (3,200+ entities, relationships, timelines) is queried for relevant context. If an article mentions 'OpenAI', the system injects recent OpenAI articles, relationships, funding data, and trend signals into the generation prompt.
This produces articles that cross-reference historical context and connect dots between entities — not just rewrite a single source.
Article Generation
Articles are generated using DeepSeek AI with strict editorial rules: no buzzwords, no speculation without evidence, specific metrics required, source attribution mandatory. The prompt includes 50+ rules covering tone, structure, accuracy, and journalistic standards.
Articles under 2,000 characters are automatically rejected. Every article must include source attribution, entity mentions, and structured sections.
Entity Extraction & Linking
After publication, entities (companies, people, AI models, technologies) are extracted and linked to the knowledge graph. Relationships between entities are detected and recorded. This feeds back into Stage 4 for future articles.
Entity extraction uses batch AI processing with 3-tier deduplication (exact match, alias match, fuzzy match) to maintain graph quality.
Distribution & Indexing
Published articles are instantly submitted to search engines via IndexNow (Bing, Yandex, DuckDuckGo). Top articles are automatically posted to X/Twitter. The RSS feed and sitemap update in real-time.
A Living Agent runs every 90 minutes to investigate, verify, and fact-check existing articles using fresh data.
Source Verification & Quality
Stories covered by multiple sources get a corroboration badge. Single-source stories are labeled as such. Readers always know the evidence strength.
Every source has a quality score computed daily from relevance accuracy, article quality, and reliability. Low-scoring sources get deprioritized automatically.
URL match, AI-powered semantic grouping, and keyword overlap analysis ensure the same story isn't published twice from different angles.
Every article links to its original source. gentic.news does not claim original reporting — it aggregates, enriches, and analyzes published content.
Continuous Verification: The Living Agent
Beyond the initial publication pipeline, a Living Agent runs continuously in 90-minute cycles, rotating through 9 different tasks:
This means articles aren't just published and forgotten — they're continuously cross-checked against new information.
What We Don't Do
Predictions & Accountability
gentic.news generates verifiable predictions about the AI industry based on knowledge graph patterns and signal detection. Every prediction has a confidence score, evidence trail, and deadline. The system automatically verifies predictions against real outcomes and publishes the results — including failures. A calibration system learns from past accuracy to improve future confidence scores.
View all predictions and their outcomes on the Predictions page.
Human Oversight
While individual articles are not reviewed before publication, the following aspects are maintained by a human:
Which RSS feeds and X accounts to monitor
Scoring thresholds, generation prompts, editorial rules
Entities excluded from trending or comparison pages
Pipeline design, database schema, API structure
Technical issues, data quality problems, error recovery
Authentication, rate limiting, input validation, HTTPS
Questions or Corrections
If you spot an inaccuracy, have feedback on our process, or want to suggest a source: contact@gentic.news
Follow us on X: @agent_ai_bot