Feed aggregator

Article URL: https://www.eventual.ai/blog/how-we-made-our-lerobot-video-reader-up-to-15x-faster

Comments URL: https://news.ycombinator.com/item?id=48939594

Points: 2

# Comments: 0

Article URL: https://wolkensteiner.substack.com/p/code-is-temporary-architecture-is

Comments URL: https://news.ycombinator.com/item?id=48939584

Points: 1

# Comments: 0

Article URL: https://artificialanalysis.ai/models/kimi-k3

Comments URL: https://news.ycombinator.com/item?id=48939580

Points: 1

# Comments: 0

Article URL: https://github.com/nodes-app/swift-markdown-engine

Comments URL: https://news.ycombinator.com/item?id=48939526

Points: 2

# Comments: 2

Article URL: https://www.tryai.dev/blog/ai-music-video-arena-claude-vs-gpt-5.6

Comments URL: https://news.ycombinator.com/item?id=48939524

Points: 2

# Comments: 1

Article URL: https://www.youtube.com/watch?v=eKelkdfi25k

Comments URL: https://news.ycombinator.com/item?id=48939523

Points: 2

# Comments: 1

Article URL: https://lemire.me/blog/2026/07/16/using-ai-to-build-your-own-software/

Comments URL: https://news.ycombinator.com/item?id=48939509

Points: 1

# Comments: 0

Article URL: https://www.popularmechanics.com/flight/airlines/a71923663/mh370-missing-plane-search-new-clue/

Comments URL: https://news.ycombinator.com/item?id=48939505

Points: 1

# Comments: 1

Here are hints and the answer for today's Wordle for July 17, No. 1,854.

Here are hints and answers for the NYT Strands puzzle for July 17, No. 866.

Here are some hints and the answers for the NYT Connections puzzle for July 17, No. 1,132.

You'll have to give your thumbs-up -- or some kind of biometric approval -- when the AI agent needs to use a password.

Over the last few weeks I've been redesigning the retrieval pipeline for an AI knowledge system.

Initially, the architecture was fairly typical:

User Question │ ▼ Vector Search │ ▼ Top K Chunks │ ▼ LLM

It worked well while the knowledge base was small.

As more documents were added, I started seeing a few recurring problems:

More irrelevant chunks being retrieved.

Larger prompts and increasing token costs.

Multiple documents discussing the same topic competing with each other.

Vector search returning semantically similar chunks from documents that weren't actually the best source of truth.

I realized the problem wasn't vector search itself.

It was deciding what should be searched before semantic retrieval even began.

Instead of treating every document equally, I separated the system into two independent stages.

Ingestion

During document upload, every document is processed once.

The pipeline extracts structured metadata including:

document type

business role

departments

topics

planner summary

retrieval keywords

authority score

importance score

answerable questions

That information is stored in a SQL registry, while document chunks and embeddings are stored separately in a vector database.

Document │ ▼ Metadata Extraction │ ├────────► SQL Registry │ └────────► Chunking + Embeddings │ ▼ Vector Store

Query Time

Instead of querying the vector database immediately, the retrieval flow became:

User Question │ ▼ Intent Analysis │ ▼ Registry Ranking │ ▼ Retrieval Planner │ ▼ Selected Documents │ ▼ Vector Search │ ▼ Context Assembly │ ▼ LLM

The registry acts as a lightweight ranking layer.

Rather than searching every document, it produces a ranked candidate set based on signals such as:

authority

importance

approval state

departments

document role

planner summary

retrieval keywords

topic overlap

The planner then decides which documents should actually participate in vector retrieval.

The vector database never searches the entire workspace anymore.

Only the planner-selected documents.

A few other changes made a noticeable difference:

similarity thresholding before accepting chunks

duplicate chunk removal

token budgeting before generation

dynamic chunk limits based on query type

ranking retrieved chunks before assembling context

One interesting observation was that improving retrieval often had a larger impact on answer quality than changing the generation model.

I'm curious whether others have moved beyond "vector search first" architectures.

If you've experimented with retrieval planning, metadata-driven routing, or hybrid retrieval systems, I'd be interested in hearing what worked and what didn't.

Comments URL: https://news.ycombinator.com/item?id=48939470

Points: 1

# Comments: 0

Recently I was baffled when Claude code read the customer table data from a production environment, while triaging an issue, and that made me wonder. Sure you should NOT give the access to read the prod data but does it sounds practical in the real time debugging session?

Comments URL: https://news.ycombinator.com/item?id=48939454

Points: 1

# Comments: 5

Article URL: https://www.reuters.com/technology/trump-media-unveils-data-feed-businesses-tracking-truth-social-posts-2026-07-16/

Comments URL: https://news.ycombinator.com/item?id=48939408

Points: 13

# Comments: 0

Article URL: https://www.nathalielawhead.com/candybox/the-wonderful-world-of-tools-made-by-small-teams-solo-devs-and-shareware-weird-beautiful-and-experimental-things-to-be-creative-in-an-analysis-on-building-for-approachability

Comments URL: https://news.ycombinator.com/item?id=48939405

Points: 2

# Comments: 0

This map shows population of each US county over time, from the US census. It also shows the US rail network over time, with data from Jeremy Atack at Vanderbilt. I think it's interesting to see the interdependence between them. And the fact that the US rail network was substantially complete by 1900.

Comments URL: https://news.ycombinator.com/item?id=48939356

Points: 1

# Comments: 0

Article URL: https://www.urbandictionary.com/define.php?term=hacker+news

Comments URL: https://news.ycombinator.com/item?id=48939346

Points: 4

# Comments: 0

Article URL: https://www.theatlantic.com/technology/2026/07/how-much-water-data-centers-use/687934/

Comments URL: https://news.ycombinator.com/item?id=48939321

Points: 1

# Comments: 0

Article URL: https://github.com/skorotkiewicz/rudo

Comments URL: https://news.ycombinator.com/item?id=48939292

Points: 1

# Comments: 0