Query verified reports, LLM research, benchmarks, and startup releases.
Enterprise AI
DiDi built a transparent, self-owned contact center quality assurance (QA) system on Amazon Bedrock, replacing an opaque third-party tool. Intent verification accuracy rose from 38...
AI Research
A collaborative paper from MIT and Stanford demonstrates that test-time search guided by lightweight formal verifiers suppresses factual hallucination in legal and medical queries ...
Open Source AI
Hugging Face released the Open-R1 pipeline, providing transparent training code, reward modeling scripts, and synthetic verification traces to replicate frontier reasoning capabili...