newsence

Discover what matters,
build your knowledge base.

Connect your sources and let AI read, summarise, and cross-reference. Every article you save compounds your knowledge.

Get started — freeSee how it works ↓
newsence.app
AllSearch articles, sources…
The Verge·

OpenAI unveils a reasoning model built for long-horizon planning

The model leads multi-step agent benchmarks and targets research and automation workflows.

AI summary · +3 links built
Hacker News·

Show HN: I rewrote an entire RSS parser in Rust

Zero-copy parsing and SIMD push throughput up 8x, with full benchmarks.

Paper

Retrieval-Augmented Reasoning: A Survey

A systematic map of recent progress joining RAG with reasoning, over 200 papers.

Stratechery·

The New Aggregation Layer

Ben Thompson on how AI becomes the new aggregation layer for content and attention.

MIT Technology Review·

Why personal knowledge management keeps failing

From Memex to today's note apps, a look at the maintenance cost nobody wants to pay.

RSS, WIRED, Hacker News, MIT Technology Review, X / Twitter, Stratechery, YouTube, TechCrunch, arXiv, LessWrong

Discover, save, create — in one place.

Stop juggling 12 tabs

The Verge·

OpenAI unveils a reasoning model built for long-horizon planning

Hacker News·

Show HN: I rewrote an entire RSS parser in Rust

Paper

Retrieval-Augmented Reasoning: A Survey

Stratechery·

The New Aggregation Layer

MIT Technology Review·

Why personal knowledge management keeps failing

Save once, never lose it

AI researchModels, papers, evals42
SemiconductorsSupply chain & process31
Daily readsToday's picks18

Ask any article

How do these three differ on reasoning models?
The split is whether planning lives inside the model, or in an external framework.1Cited 3 sources

Turn reading into output

Reasoning models weekly: three schools of thought

Rewrite this — tighter

The open web is the richest library humanity has ever built. It is also the most scattered.

Articles, threads, papers, videos — the pieces worth keeping arrive in a dozen apps, written by a hundred thousand people, on a schedule no one controls. Browsers were built for visiting, not keeping: we bookmark and forget, save and never find.

There has to be a better way to hold onto what you read.

Two old ideas, one new tool.

1945
The Memex desk plate from Life magazine, September 1945

Vannevar Bush — Memex ↗

Bush imagined a personal, curated store with associative trails between documents. The piece he could not solve: who does the maintenance.

2026
LLM AGENT in the form of a background process reads what you save,files it, links people and ideas, and keeps the wiki growing by itself.

Andrej Karpathy — LLM Wiki ↗

Karpathy named the modern answer: an LLM agent that quietly maintains a compounding wiki while you read. The bookkeeping is the LLM's job.

newsence is the new tool.

Start free. Upgrade when ready.

Free is enough to start. Pro is where you pay to keep your knowledge on your side.

Explorer

Start free. Everything public by default — nothing hidden.

$0forever
Get started — free
  • —Unlimited public documents
  • —500 AI credits / month
  • —Collections & web clipper

Pro

Private by default. Your knowledge vault, not ours.

$15/month
Upgrade to Pro
  • —Everything in Explorer
  • —Private documents & collections
  • —15,000 AI credits / month
  • —Web search & image generation

Stop letting what you read
vanish into the next tab.

Connect your sources and see your first AI summary within five minutes.

Start building your knowledge base — free