Tag
AstraZeneca describes Research Assistant, an internal LLM-based multi-agent system that lets scientists explore biomedical data via chat, with evidence grounding and citation links, deployed to 15,000 internal users.
Specula is an agentic system that automatically derives TLA+ specifications from code, runs model checking to find concurrency bugs, and reproduces them via integration tests. It found 249 bugs across 48 open-source distributed and concurrent systems, demonstrating a significant scaling of formal verification.
Meta presents Shape Your Feed (SYF), an LLM-based agentic framework for real-time conversational recommendation that co-curates content via multimodal input, agentic re-ranking, and self-evolution with DPO, achieving strong offline and online results.
This paper introduces PIMiner, an agentic system for automatic prompt injection red-teaming that builds a strategy library during training and transfers to unseen target LLMs at test time, achieving strong attack success rates against models like Gemini-2.5-Pro, GPT-5.1, and Claude-Sonnet-4.5.
TigrimOSR v0.7.2 is an open graph agentic system in Rust, adding a lightweight CLI mode (~4 MB RAM), prebuilt binaries for macOS/Windows/Linux, and YAML-based configurability for agent graphs, models, and tools.
This paper introduces MEDA, an LLM- and symbolic regression-powered agentic framework for automatically discovering ordinary differential equation (ODE) models of biological dynamical systems. It retrieves background knowledge, proposes candidate ODEs, and evaluates them across canonical model retrieval, extrapolation, and open-ended discovery tasks, demonstrating strong structural recovery and biologically plausible models.
A senior Anthropic engineer published a playbook for building autonomous agentic loops with components like Generator, Evaluator, Memory, and Orchestrator, claiming it reduced his workload by 95% and doubled the firm's revenue.
Sakana AI has released AI Scientist-v2, an end-to-end automated research system that can autonomously go from generating research hypotheses to writing papers, and has been accepted by the ICLR2025 Workshop after peer review.
RocketSmith is an agentic system that uses large language models to automate the design and additive manufacturing of high-powered rockets, achieving successful flight tests with simulation results matching 80% of predicted apogee.
OpenMontage is the world's first open-source Agentic video production system. With just one sentence, it automates the entire workflow from research and scripting to dubbing and compositing. It supports 12 major Pipelines and 52 tools, with costs as low as a few cents. It has over 16k stars on GitHub.
A 22-chapter skeleton course on building production AI agents, using an innovative approach where the AI partner fills in details. The course covers tool calling, agent loops, memory, multi-agent collaboration, and more.
Introduces ModSleuth, an agentic system that recursively reconstructs large-scale dependency graphs for LLM development by analyzing public artifacts, revealing multi-hop license obligations and documentation inconsistencies.
A guide on building a secure agentic system with sandboxing, parallel sub-agents, tool calling with control policies, inference routing, and protection against injection and role escalation attacks, to be published by Evangelos Pappas.
Introduces Benchmark Agent, a fully autonomous system for creating diverse benchmarks with minimal human intervention, enabling continuous model assessment across domains.
Adaptive Auto-Harness is a framework for sustained self-improvement of agentic systems deployed on open-ended task streams, outperforming baselines via a stateful multi-agent evolver, harness tree, and human-steering hooks.
A developer shares their experience building a local autonomous agent with long-term and short-term memory using Qwen 3.6 27B, noting that memory dramatically improves the agent's usefulness and realism. They invite others building similar agents to discuss memory techniques and potential agentic meetups.
The team of Assistant Professor Elliott Wu from the University of Cambridge released Articraft — an agentic coding system that can automatically generate interactive 3D assets with parts, joints, and realistic motion, and open-sourced the Articraft-10K dataset, significantly lowering the asset threshold for robot training and physical AI.
Microsoft's new multi-model agentic security system (MDASH) tops the CyberGym leaderboard for vulnerability discovery, achieving 35 zero-day findings and demonstrating advanced AI-driven defense capabilities.
The author shares a locally runnable AI companion built with Python, Gemini, and Ollama, featuring a custom cognitive architecture based on Global Workspace Theory and an Integrated Information Theory proxy for personality modeling.