HACKOBAR_
HACKOBAR_ is one feed for AI news - a ranked digest from HN, GitHub, arXiv, Reddit, X, lab blogs, and newsletters.
Last verified:
What is HACKOBAR_?
HACKOBAR_ is an AI news aggregator that consolidates AI-related content into a single, swipeable feed of 25 top cards updated every 30 minutes. It fetches news from Hacker News, arXiv, GitHub trending, HuggingFace, AI subreddits, AI-focused Twitter handles, tech outlets, AI lab blogs, and newsletters, then runs a 5-step pipeline (keyword + LLM relevance filtering, three-layer deduplication, engagement/recency scoring, Claude-powered 50-word summarization) to surface the best content. Designed for engineers, researchers, builders, founders, and investors; no login or paywall.
HACKOBAR_ pricing
Pricing model: Freemium
Hackobar is completely free to use with no paywall and no login required. The free tier includes full access to the AI news feed with 25 cards updated every 30 minutes, all filtering/deduplication/scoring features, Claude-powered summaries, and the role-based lens feature. The project initially ran on Cloudflare's free tier but shifted to a paid infrastructure model to handle complex deduplication. There is no paid Pro tier mentioned for the AI news aggregator - the $1-5/month Pro pricing mentioned in some sources appears to reference a different Hackobar product (a streak-based builder log for solo developers).
HACKOBAR_ pros
- Aggregates AI news from 10+ sources in one feed
- No login required to use the service
- No paywall - completely free to access
- 25-card swipeable feed for easy navigation
- Updates every 30 minutes for fresh content
- AI-powered deduplication prevents duplicate stories
- Claude-generated 50-word hype-free summaries
- Focuses on architecture, benchmarks, and real impact
- Role-based lens for builders/researchers/founders/investors
- Source weighting prevents research from being buried by social noise
- Original article links available for deep reading
- Cross-platform scoring bumps original sources
- Recency decay in scoring keeps feed balanced
- Hype-free summaries concentrate on facts
- Next.js stack ensures fast page loading
HACKOBAR_ cons
- Only shows 25 items at a time
- Updated every 30 minutes (not real-time)
- May miss niche AI sources not in pipeline
- Summary length limited to 50 words
- Role-based lens usefulness is debatable
- Depends on Claude API which may have costs
- No mobile app available yet
- No custom source addition by users
Frequently asked questions about HACKOBAR_
What sources does Hackobar aggregate from?
Hackobar fetches AI news from Hacker News, arXiv, GitHub trending repositories, HuggingFace, many AI subreddits, Twitter handles that cover AI, tech news outlets, major AI lab blogs, and numerous AI newsletters that are growing by day.
How often is the feed updated?
The feed contains 25 items and is updated every 30 minutes, ensuring you get fresh AI news without having to constantly check multiple sources.
Do I need to create an account to use Hackobar?
No, Hackobar requires no login at all. You can access the entire feed immediately without creating an account or providing any personal information.
Is Hackobar free or does it have a paywall?
Hackobar is completely free with no paywall. All features including the news feed, summaries, filtering, and role-based lens are accessible without any payment.
How does Hackobar filter out non-AI content?
Hackobar uses a two-step filtering process: first filtering articles for specific top-level keywords, then using Gemma 4 26B batch classifier to determine if each item is AI-relevant or not.
How does Hackobar handle duplicate articles across sources?
Hackobar uses a three-layer deduplication pipeline: URL normalization, Jaccard similarity on word tokens, and Gemma semantic matching. When a duplicate is found, the original news item's cross-platform score gets bumped instead of dropping the duplicate.
How are articles scored and ranked in the feed?
Each item is scored based on engagement metrics, LLM signal, cross-platform signal indicating how many sources covered it, and recency decay. Each source also has a weight component that keeps research from getting buried by social media noise.
What technology powers Hackobar?
Hackobar uses Next.js on Cloudflare Pages, Hono API worker, Supabase for the database, Anthropic Claude for generating summaries, and Gemma via Cloudflare Workers AI for classification and semantic matching.
What is the role-based lens feature?
The lens feature filters the entire feed and highlights how each AI news item specifically impacts different roles: builders, researchers, founders, investors, policy makers, and healthcare workers, helping you see news relevant to your profession.
Can I read the original articles or just the summaries?
Yes, while Hackobar provides Claude-generated 50-word summaries for quick scanning, each card includes a 'details' link that takes you to the original article if you want to read the full content.