Loading activity…
@cshorten30
AI and Databases @weaviate_io
AI
The reported MATH-500 score rose from 7.9% to 48.1% with the same sampling settings.
AI
Weaviate says MIT doctoral student Jacob X. Li discusses AI agents studying code or logs before knowing their tasks.
AI
Nathan Lambert's slides distinguish open weights from open source and examine competition with proprietary models.
YouTubeAI
A selected team says it aims to improve large-scale AI-powered data processing and deploy its work with industry partners.
AI
Laude Institute says SkyDiscover can also assemble complete systems from a specification.
AI
The Weaviate Podcast description says Başak Eskili discusses a central MCP server for internal APIs and external tools.
AI
A post celebrating the milestone describes clips of conversations with people building AI-native apps, databases and search.
AI
Booking.com's ML & DS blog says its tests used 100 million embeddings, filtered searches and concurrent writes.
AI
Adaption AI says it released the report while asking how to post-train a model for a new capability with zero data.
AI
A user calls work-update emails from their agent a “massive unlock.” They say they currently use Claude Cowork with Gmail’s interface after every scheduled task.
AI
Cohere lists Pro at $0.12 and Fast at $0.08 per million tokens. It says teams can mix the tiers without rebuilding their index.
AI
A user testing a JevRAG for exploring papers says they’ve had more success using Jev for reranking and found interesting ways to pair it with semantic search.
AI
A June 2026 post introducing the episode says Allen discusses his experience in the expert-systems era and topics including knowledge graphs, LLM-based classifiers and theorem provers.
AI
The Weaviate Podcast's host says the goal isn't a more human-sounding assistant, but a model of behavior that includes pushback, confusion and frustration.
AI
An announcement for Weaviate Podcast episode 144 highlights Alex Ledbetter of Delegance Brokerage, citing 900 pull requests shipped in three months with 18 terminal agents running.
AI
An author of “Querying Databases with Function Calling” says models often choose LIKE or exact text matching for natural-language commands that require SEARCH.
AI
The author reports that models translating natural-language commands into Weaviate queries often use LIKE or exact-match operators for commands requiring the SEARCH operator.
AI
Weaviate Podcast describes a benchmark built by Shreya Shankar and colleagues with 54 queries spanning 12 datasets and four database systems, testing how AI agents answer questions across databases.
AI
A post sharing a Weaviate Podcast episode says the key lesson was giving an analysis agent tools to investigate why relevant documents ranked poorly, rather than relying on a score alone.
AI
Connor Shorten shares highlights from Akari Asai's VLDB talk on DR Tulu and AgentIR.
