Radar

Curated AI research, releases, and signals — filtered for what matters to people building with AI.

This Week

arXivFeb 25

New policy optimization method solves training collapse in multi-step agentic RL tasks — more reliable agent training.

AI Agents

TechCrunchFeb 24

Opal now lets you build multi-tool agentic workflows with persistent memory — no-code agent orchestration goes mainstream.

AutomationAI Agents

HuggingFace PapersFeb 23

Cross-modal RAG framework that integrates docs, images, and tables — relevant for building knowledge layers in custom systems.

Applied AILLM Infra

Earlier

AnthropicFeb 17

Opus-class coding and agent performance at Sonnet pricing — drops the cost barrier for production agentic workflows.

IndustryAI AgentsLLM Infra

OpenAIFeb 5

OpenAI's most capable agentic coding model — 25% faster, new SWE-Bench highs, signals where coding agents are heading.

IndustryAI Agents

Mistral AIFeb 4

Sub-200ms streaming transcription with diarization, open weights — enables real-time meeting intelligence in custom systems.

LLM InfraMulti-Model

AnthropicJan 21

Developers integrate AI into 60% of work but fully delegate only 0-20% — the shift from writing code to orchestrating agents.

AI AgentsIndustry