Towards a decentralized and collaborative search engine
summary
A collaborative search engine called all.site combines curated, community-run indices with ML question answering, using Hugging Face QA models over Elasticsearch/OpenSearch results; admins pick trusted sources, pin results, and publish config changes like GitHub. It surfaces version-aware docs (Solr/Elasticsearch) and indexes non-web sources via a Slack connector, cites sources in generated answers, and runs as a technical preview. About a quarter of attendees already use ChatGPT-style tools, and the prototype is live but still tackling duplication, scalability and security.
AI-written summary · about
chapters (10)
- 0:00Intro
- 0:38Questions
- 2:54Ads
- 10:33Alternative
- 15:31Experts
- 20:22Question Answering
- 28:42Slack
- 29:32Data curation
- 31:04Transparency
- 32:40Technical challenges
related talks
10mDecentralized Search Over Decentralized Storage
25mIt's Time to Reinvent the Web, Without the Internet
33mAI Your Data
33mLocal-first and Sync Engines, the Architecture of the Future?
24mMonetizing Open Source via the Decentralized Cloud Stack
28mOscoin: A P2P Alternative for OSS Collaboration
16mPlanetServe: A Decentralized, Scalable, and Privacy-Preserving Overlay for...
41mCreating Local-First Collaboration Software with Automerge