Sycamore – an LLM-powered semantic data preparation system for search
We’re Aryn and yesterday we came out of stealth: blog.aryn.ai. As part of that, we released Sycamore: https://github.com/aryn-ai/sycamore. Sycamore is an LLM-powered semantic data preparation system for building search applications. It introduces a distributed set-based abstraction, a DocSet, that makes processing a large document collection as easy as reading a single document. Sycamore makes it easy to use LLMs to transform and enrich your unstructured data and prepare it for search. It comes with a scalable distributed runtime, built on Ray, that makes it easy to go…
In plain words
Sycamore is an LLM-powered system for preparing unstructured data to build search applications. It introduces a DocSet abstraction that simplifies processing large document collections, allowing developers to transform and enrich data with a few lines of code. The system includes a distributed runtime built on Ray for scaling from prototypes to production. It's designed for developers building search applications who need to process documents like PDFs, extract entities, compute embeddings, and load results into search platforms.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
We’re Aryn and yesterday we came out of stealth: blog.aryn.ai. As part of that, we released Sycamore: https://github.com/aryn-ai/sycamore. Sycamore is an LLM-powered semantic data preparation system for building search applications. It introduces a distributed set-based abstraction, a DocSet, that makes processing a large document collection as easy as reading a single document. Sycamore makes it easy to use LLMs to transform and enrich your unstructured data and prepare it for search. It comes with a scalable distributed runtime, built on Ray, that makes it easy to go from prototype to production. For example, with Sycamore, you can read a collection of PDFs, partition them in coherent chunks, pull out entities like titles and authors, compute vector embeddings, and load them into a local OpenSearch cluster. All with a few lines of code. To learn more, visit the repo: https://github.com/aryn-ai/sycamore, docs: https://sycamore.readthedocs.io/, and demo: https://www.loom.com/share/53e68b0eb5ab49948111a3fcf6286b7f?... We’d love for you to try it out, give us feedback, and contribute.
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
