CaseCite is a self-hosted legal research and document-analysis platform designed for firms that want to keep their document corpus and application data on their own infrastructure. It combines eight tools around one shared knowledge base: question-answering over firm documents with source citations; document uploads and cloud-storage integrations; contract review, comparison, and redlining; drafting grounded in selected references; chat with CourtListener opinions; citation mapping; judge analytics; and case-law lookup tools. A user-defined practice profile supplies context across reviews and drafts. Long-document drafting uses a section plan and separate drafting windows before reconciling the result. The project emphasizes that AI outputs need verification and that its citation check is not a substitute for commercial citators such as Shepard’s or KeyCite.
The repository provides Docker-based deployment instructions and lets firms configure their own AI provider keys, including OpenAI, Anthropic, or Gemini for language tasks. Documents, indexes, chat history, and analyses remain on the deployment, but using AI sends relevant text to the configured providers; uploads also send text to an embeddings provider, while case-law searches go to CourtListener. Fully local processing requires self-hosted chat and embedding models. Uploaded documents and API keys receive application-layer encryption, but derived text such as vector chunks and chat history is stored unencrypted, making host-level disk encryption the operator’s responsibility. CaseCite is source-available under the Elastic License 2.0: internal self-hosting and modification are permitted, but offering it as a hosted service is not.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Google AI Mode sometimes avoids directly saying Joe Biden won the 2020 US presidential election, instead directing users to official sources. Its responses vary across users and election years, while answers become more direct when asked outside the US. (arstechnica.com)
AI data centers can consume millions of gallons of water daily, competing with local communities for drinking water as their numbers grow. Residents around the world are organizing to oppose projects that threaten local water supplies. (restofworld.org)
Cartesia argues that tokenizers limit end-to-end learning because they impose hand-crafted text boundaries. Its hierarchical networks learn to group raw bytes into useful units, offering a scalable path toward byte-level language models. (cartesia.ai)
Particle-1 is a software-engineering model post-trained on DeepSeek V4.1 Flash that uses about 45% fewer tokens while maintaining comparable accuracy. Across four coding and data-engineering benchmarks, it cuts average task costs by 34.4%-51.6%, with plans to expand into other domains. (particle.ai)
Google’s Gemini gave different results for two election-related queries that differed by one digit, according to an X post. The post claims one result censored Joe Biden but does not show the queries or responses. (twitter.com)
Google introduced Gemini 4 Carbon, a new model in its Gemini lineup. The announcement describes the model and its capabilities. (blog.google)
Today's best Hacker News stories, summarized and screenshotted, one email a day.