> about
The AI Wire is a curated feed of AI research, models, papers, tools, and comparative analysis — updated daily from a systematic research pipeline.
How it works
Every day, an automated pipeline scans sources including arXiv, HuggingFace, GitHub, and research aggregators to identify notable AI developments. Each item is categorized, scored for relevance, tagged, and stored with full context.
Categories
- Models — New LLMs, vision models, multimodal systems, and frontier releases
- Papers — Research papers and preprints across ML, NLP, CV, and related fields
- Tools — Infrastructure, evaluation platforms, frameworks, and developer tools
- Comparisons — Head-to-head analysis of models, tools, and approaches
Relevance scoring
Each item is scored 1-5 for relevance based on novelty, impact, and practical applicability. The green dots next to each article indicate its relevance score.
Data sources
Research is collected via Perplexity deep research, arXiv API, HuggingFace Hub, and manual curation. The pipeline has been running daily since November 2025, covering 100+ days of continuous AI landscape monitoring.
Technology
Built with Next.js, SQLite (FTS5), and Tailwind CSS. Data is stored in markdown, parsed into a SQLite database with full-text search, and served as a static-first site with incremental regeneration.
Created by Justin Johnson. Powered by systematic curiosity.