Collect
Scheduled ingest pipelines per track and source-local-first, SQLite-backed.
Vtory · PegasusTech · Developed 2026
NewsIntelligence is a product of Vtory and PegasusTech, developed in 2026. AI agents collect headlines from RSS, Hacker News, Reddit, GitHub, arXiv, YouTube, and more; classify each item against your tracks; and surface what is worth your time.
Collect
Scheduled ingest pipelines per track and source-local-first, SQLite-backed.
Classify
LLM labeling with track-specific questions, keep/skip rules, and ranked scores.
Deliver
Explorer UI, REST API, and MCP tools so humans and agents query the same signal.
Whether you follow AI policy, e-commerce platforms, or your own niche, you define a track-sources plus classifier prompts. The system does the rest. Shipped as a joint Vtory & PegasusTech product line, actively developed since 2026.
Plug in feeds and search adapters. Items land in one store with provenance, ready for labeling and search.
Each article is scored against your track: relevance, custom labels, and a plain-language reason-so “keep” means keep for you.
MCP exposes list_tracks, search_news, and get_item so Cursor and cloud agents pull classified news into workflows.
Built by the Vtory & PegasusTech team on Anthropic’s stack for reasoning-heavy steps-classification, assistant flows, and agent integrations-not generic chat wrappers.
Open source: github.com/PegasusTech-VN/NewsIntelligence
Leave your work email for early access, hosted trials, or integration questions. We typically reply within one to two business days.
A news intelligence platform from Vtory and PegasusTech (2026): ingest from many sources, LLM classification per track, Explorer UI, REST API, and MCP for AI agents. It replaces ad-hoc RSS tabs and manual skimming with structured, queryable signal.
Founders, analysts, and operator teams who need domain-specific news-e-commerce, AI, compliance, or custom tracks-without reading everything raw.
Clone the GitHub repo, configure sources and tracks, run ingest + label jobs, and optionally expose MCP to your agent stack. See the README for CLI and environment variables.