frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

I want to help automate your business insights

https://www.trynexus.io/
1•nikpil•7m ago•1 comments

Humans used to have straighter teeth – what changed?

https://www.nationalgeographic.com/health/article/crooked-teeth-human-evolution-jaw-size
1•bookofjoe•9m ago•1 comments

The Silicon Valley push to breed super-babies

https://www.washingtonpost.com/technology/2025/07/16/orchid-polygenic-screening-embryos-fertility/
1•pabs3•9m ago•0 comments

Running NetBSD on my Amiga 4000

http://sandervanderburg.blogspot.com/2025/02/running-netbsd-on-my-amiga-4000.html
1•doener•10m ago•0 comments

Running Linux on my Amiga 4000

http://sandervanderburg.blogspot.com/2025/01/running-linux-on-my-amiga-4000.html
1•doener•10m ago•0 comments

Conferences, Clarity, and Smokescreens

https://infrequently.org/2025/06/conferences-clarity-and-smokescreens/
1•pragmatic•11m ago•0 comments

What everyone gets wrong about AI customer support

https://www.mux.com/blog/all-the-wrong-ways-to-think-about-ai-customer-support
1•dylanjha•14m ago•0 comments

No vegan milk is equivalent to dairy, nutritionists conclude

https://www.thetimes.com/life-style/health-fitness/article/plant-based-milk-cows-nutrition-ct7dlbxgl
1•josephcsible•15m ago•0 comments

Using Typst to Typeset Novels

https://splinterton.com/blog/typst_notes/
2•raybb•18m ago•0 comments

Pair-Instability Supernova

https://en.wikipedia.org/wiki/Pair-instability_supernova
1•dustingetz•19m ago•0 comments

Reflections from Toxic Engineering Teams

https://resync-games.com/blog/engineering/toxic-teams
1•kadhirvelm•21m ago•1 comments

Mammals Evolved into Ant Eaters 12 Times Since Dinosaur Age, Study Finds

https://news.njit.edu/mammals-evolved-ant-eaters-12-times-dinosaur-age-study-finds
2•zdw•21m ago•0 comments

23andMe is out of bankruptcy. You should still delete your DNA

https://www.washingtonpost.com/technology/2025/07/17/23andme-bankruptcy-privacy/
2•1vuio0pswjnm7•21m ago•0 comments

Im using p5.js in production for a simplistic video/type editor

https://www.typecut.xyz/
1•ghostuniversity•23m ago•0 comments

Accelerate Generative AI Inference with Nvidia Dynamo and Amazon EKS

https://aws.amazon.com/blogs/machine-learning/accelerate-generative-ai-inference-with-nvidia-dynamo-and-amazon-eks/
1•dtap•23m ago•0 comments

Robots in China are riding the subway to make 7-Eleven deliveries

https://www.popsci.com/technology/robots-in-china-subway-7-eleven-deliveries/
2•domofutu•24m ago•0 comments

Ask HN: (Semi) Automated LinkedIn Connection Reviewer

1•bix6•25m ago•0 comments

Colombia Bets Big on Instant Payments with Central Bank's BRE-B

https://thecitypaperbogota.com/business/colombia-bets-big-on-instant-payments-with-central-banks-bre-b/
1•raybb•31m ago•0 comments

Amazon EKS ultra scale clusters

https://aws.amazon.com/blogs/containers/under-the-hood-amazon-eks-ultra-scale-clusters/
2•ra7•32m ago•0 comments

Nuxt 4.0

https://nuxt.com/blog/v4
2•CharlesW•32m ago•0 comments

Self Defence With Steve Powell (1982) [video]

https://www.youtube.com/watch?v=miYXEzG9huc
1•austinallegro•36m ago•0 comments

Ask HN: Have You Ever Rediscovered a Book You Forgot You'd Read?

2•bicepjai•37m ago•0 comments

People kept working, became healthier while on basic income: report

https://www.cbc.ca/news/canada/hamilton/basic-income-mcmaster-report-1.5485729
10•jszymborski•38m ago•0 comments

Techniques scraping vulnerability data with over 100 crawlers (without LLMs)

https://secalerts.co/news/techniques-scraping-vulnerability-data-from-100-different-sources-without-llms/1iVmylCNLzLCdgOsnpkW34
1•louisstow•43m ago•1 comments

Two Simple Rules to Fix Code Reviews

https://serce.me/posts/2025-07-17-two-simple-rules-to-fix-code-reviews
1•SerCe•44m ago•0 comments

DOC Study on Economic Benefits of GPS (2019)

https://space.commerce.gov/doc-study-on-economic-benefits-of-gps/
1•teleforce•45m ago•0 comments

Show HN: Groolp, a task automator written in Go

https://github.com/ystepanoff/groolp
1•ystepanoff•47m ago•0 comments

Netflix is entering a new era of growth

https://www.businessinsider.com/netflix-earnings-preview-wall-street-password-sharing-ad-tier-2025-7
1•mgh2•49m ago•2 comments

Delivery of biologics via the GI tract by self-pressurized oral capsules

https://www.sciencedirect.com/science/article/pii/S0168365925005838
1•PaulHoule•50m ago•0 comments

Show HN: Default Avatar Generator

https://github.com/pas256/default-avatar-generator
1•pas256•51m ago•0 comments
Open in hackernews

Achieving SOTA LongMemEval scores (80%) with RAG alone

https://mastra.ai/blog/use-rag-for-agent-memory
3•tybaa•3h ago

Comments

tybaa•3h ago
At mastra.ai we ran the LongMemEval benchmark (500 questions across thousands of conversations) to systematically test our agent memory features. After seeing claims that "RAG is dead for agent memory", we decided to see what was possible.

Starting at a low 65% accuracy, we made some changes to how our memory system works and reached 80% using RAG alone. We ran the benchmark with a series of different configs (since we're a configurable framework) and saw results ranging from 63% with very conservative settings, 74% with small to medium context size, up to 80% with longer context. We accidentally spent $8k and burned 3.8B tokens figuring this out - but it proved that RAG absolutely works for agent memory when properly configured.