frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Semantics Delivery Network: Rethinking Web Retrieval for LLM Agents

https://arxiv.org/abs/2609.22486
1•scottsiume•51m ago

Comments

scottsiume•51m ago
When using an AI agent to look up information, we usually just dump an entire webpage, website, or directory into the model and let it "brute force" its way through to pick out a few potentially useful paragraphs.

This paper proposes a different approach:

1. Use a query retrieval system to push only the relevant paragraphs to the AI. 2. Consolidate and deduplicate the information from those sections.

This approach cuts costs and runs much faster.

This method solves the efficiency problem, but accuracy cannot be guaranteed. It relies entirely on the capability of the query system mentioned in the first point here.

Another way to improve data accuracy is to extract and transform the semantic layer before data entry. This does make writing data slower, but semantic data is different from transactional business data.

Semantic data doesn't necessarily need to be consumed by operational systems immediately. From a database perspective, it doesn't require OLTP characteristics. Therefore, a better approach might be to let this data undergo a period of processing and cleaning to become meaningful semantic information before making it available for subsequent retrieval.

Tales from the Software Factory: The Cron and the VM

https://blog.exe.dev/cron-vs-vm
1•vinhnx•1m ago•0 comments

I built a language model out of cellular automata

https://ssenthilnathan3.github.io/blog/cellular-automata-language-model/
1•nathaah3•3m ago•0 comments

Update: Source Download – Inspect, grab assets, screenshot and record video/GIF

https://chromewebstore.google.com/detail/source-download/nockdgincmpfojabnhbofkddgcmnodpd
1•2rundev•4m ago•0 comments

Omens of Exceptional Talent

https://guzey.com/talent/
1•karakoram•4m ago•0 comments

Debian Inference Portal (Unofficial)

https://inference.debian.net/
1•selfhoster1312•5m ago•0 comments

Understanding Runtime Governance for Enterprise AI Deployment

1•thespotlessmind•7m ago•0 comments

Investigate top humanity questions with your own AI agent

https://interworky.com/
1•ahmedhussam1053•7m ago•1 comments

Can open-source prompt-injection detectors catch realistic AI agent attacks?

https://github.com/rudratoshs/buried-injections
2•northbridgedev•9m ago•0 comments

AI Passport Photo

https://www.aipassportphoto.org/
2•wsm123456•11m ago•1 comments

How to Audit an AI Agent:Time-Travel Debugging and Drift Measurement

https://medium.com/@MirArshadTalpur/how-to-audit-not-observe-an-ai-agent-time-travel-debugging-an...
1•Arshad-Talpur•12m ago•0 comments

The cost and quality frontier across 29 AI coding models

https://bito.ai/benchmarks/ai-coding-model-cost/
1•kirti_soni171•12m ago•0 comments

Inferact's Kimi K3 Megakernel hits 700 tok/s on TPU, Beating GPU

https://inferact.ai/blog/tpu-megakernels
4•xutingl•13m ago•1 comments

Europe's fastest supercomputer is finally getting its domestic silicon infusion

https://www.theregister.com/systems/2026/09/23/europes-fastest-supercomputer-is-finally-getting-i...
1•doener•14m ago•0 comments

FBI investigating hacking group's claim of breach of agent info

https://www.nbcnews.com/tech/security/fbi-investigating-hacking-groups-claim-massive-breach-agent...
1•ycombinatronics•14m ago•0 comments

Dolphin Progress Release 2609

https://dolphin-emu.org/blog/2026/09/24/dolphin-progress-report-release-2609/
1•madiling•16m ago•0 comments

Show HN: I asked 5 LLMs 10k questions and indexed every domain they cited

https://citegeo.uvansa.com
1•jhaankit373•18m ago•0 comments

Meta ditches the camera on its newest smart glasses

https://www.theverge.com/tech/999281/ray-ban-meta-audio-glasses-meta-connect-2026
4•geoffbp•21m ago•0 comments

The E-Ink Monopoly [video]

https://www.youtube.com/watch?v=3QbJ7pnc1Lc
1•trymas•24m ago•0 comments

Blink – Embeddable one-pass typed decision in 33 µs

https://github.com/sqliteai/blink
1•marcobambini•25m ago•0 comments

What Changed from GPT-3.5 to GPT-4? GPT-3.5: 0/30 Empty Nulls. GPT-4: 30/30

https://www.reddit.com/r/ArtificialInteligence/comments/1wotljm/what_changed_from_gpt35_to_gpt4_s...
2•rayanpal_•26m ago•0 comments

Steve Jobs' Full iPhone 4 Antennagate Press Conference Q&A Session

https://www.youtube.com/watch?v=BiN5ERktXz0
4•aSig•28m ago•0 comments

Can You Beat Jev?

https://antics.gg/can-you-beat-jev
1•thatslast•30m ago•0 comments

Albanese says OpenAI hacked Medicare and told Australia months later

https://www.theguardian.com/australia-news/2026/sep/24/anthony-albanese-says-openai-agent-hacked-...
1•skor•32m ago•0 comments

What's Left to Get Wrong

https://blogsh.app/posts/2026/what-s-left-to-get-wrong/
1•krab•33m ago•0 comments

Meta is making a standalone Muse AI gadget

https://www.theverge.com/tech/999750/muse-charm-meta-ai-hardware
1•arbayi•33m ago•0 comments

Mysteries of AI Generalization

https://www.astralcodexten.com/p/mysteries-of-ai-generalization
1•terryf•33m ago•0 comments

Lysios – LLM red teaming org

https://lysios.ai/
2•davedx•34m ago•0 comments

The biggest surprises from a year working at OpenAI

https://aalokpatwa.substack.com/p/one-year-at-openai
1•mji•35m ago•0 comments

Ways to Encrypt Data on Servers

https://lwn.net/Articles/1092553/
1•DavideNL•38m ago•0 comments

C-to-Rust Fallacy: Automatic Refactoring != Memory Security

https://arxiv.org/abs/2609.25682
3•sbulaev•38m ago•0 comments