Agents

Save to Spotify is a new command-line tool designed specifically for AI agents like OpenClaw, Claude Code, or OpenAI Codex. If you're the kind of person who collects research on a topic, then feeds it through their AI of choice to create audio summaries and personal podcasts, this lets you save them right alongside the latest episode of The Vergecast and Welcome to Night Vale on Spotify. To set it up, you need to download and install the Save to Spotify CLI from GitHub. Then you just prompt your AI agent as normal, but tack on "and save to Spotify," and it should show up right in your podcast...

Parloa uses OpenAI models to build voice-driven customer service agents with simulation and real-time deployment capabilities for enterprises.

LongSeeker proposes Context-ReAct paradigm for elastic context management in long-horizon search agents, maintaining trajectory at variable detail levels.

Coding agent with executable Python world models, verification, and simplicity-bias refactoring solves 25 public ARC-AGI-3 games without task-specific logic.

Also, rate limits will double for Pro and Max users of tools like Claude Code.

Long-horizon LLM agents depend on intermediate information-gathering turns, yet training feedback is usually observed only at the final answer, because process-level rewards require high-quality human annotation. Existing turn-level shaping methods reward turns that increase the likelihood of a gold answer, but they require answer supervision or stable task-specific verifiers. Conversely, label-free RL methods extract self-signals from output distributions, but mainly at the answer or trajectory level and therefore cannot assign credit to intermediate turns. We propose Self-Induced Outcome Po...

Anthropic launches 10 finance-focused AI agents via Claude Cowork and Managed Agents for KYC screening, pitchbook generation, and month-end close workflows.

SAP plans to buy German AI startup Prior Labs and invest heavily in it. It is also prohibiting customers' agents use to a select few like Nvidia's NemoClaw.

OpenSeeker-v2: SFT on informative trajectories achieves frontier LLM search agent capabilities without full RL pipeline.

MOSAIC-Bench evaluates coding agents' vulnerability to multi-stage attack chains that decompose malicious goals into innocuous sequential tasks, exposing alignment gaps in deployed systems.

Anthropic releases ten Cowork and Claude Code plugins plus Microsoft 365 integrations and MCP app for financial services.

The automotive cockpit is undergoing a fundamental shift from rule-based interfaces to agentic, multimodal AI systems capable of reasoning, planning, and... The automotive cockpit is undergoing a fundamental shift from rule-based interfaces to agentic, multimodal AI systems capable of reasoning, planning, and acting. In most vehicles on the road today, in-vehicle assistants still rely on fixed command-response patterns: interpret a phrase, trigger an action, reset. While effective for well-defined tasks, this approach doesn’t scale to modern… Source

Argues frontier AI failures in open-ended tasks (scientific assistance, agents, personalization) stem from objective ambiguity rather than capability gaps; proposes contextual multi-objective optimization.

Generative AI’s explosive first chapter was defined by humans sending requests and models responding. The agentic chapter is different. Agents don't... Generative AI’s explosive first chapter was defined by humans sending requests and models responding. The agentic chapter is different. Agents don’t follow a pre-determined sequence of actions. They call tools, spawn sub-agents with different tasks and models, retain information in memory, manage their own context window, and decide for themselves when they’re finished. In doing so… Source

ProgramBench: 200-task evaluation showing agents struggle to rebuild large binaries from scratch without cheating vulnerabilities.

The Seattle-based startup's Series A round was led by Glilot Capital, NFX and SignalFire, TechCrunch has exclusively learned.

OpenAI and PwC partner to deploy AI agents for enterprise finance automation, forecasting, and CFO workflow modernization.

FlexSQL agent flexibly explores schemas and data during text-to-SQL generation, enabling recovery from early mistakes.

Opinion piece argues LLM agents require jointly formulated actions and plans with human actors rather than isolated architectures.

ORPilot: open-source agentic system translating ambiguous business problems into solver-ready optimization models with conversational and data collection agents.

Zero-trust authorization framework for LLM agents with hybrid inspection and task-based access control to mitigate tool-use and resource-access risks.

Novel framework for bandwidth-efficient remote control via minimal information transmission between controller and agents in continuous action spaces.

DataEvolver implements closed-loop agent-driven visual data generation and refinement for image editing, supporting masks, depth, poses, and trajectory artifacts.

Momentum integrates runtime procedural content generation and autonomous agent evaluation in endless-runner gameplay to assess generated terrain balance and solvability.

Iterated negotiation benchmark tests LLM agents' ability to repair grounding failures in dynamic multi-turn interaction.

GRAVITY module injects relational, temporal, and thematic structure into conversational memory retrieval for long-horizon agents.

AutoMat benchmark evaluates LLM agents on reproducing computational materials science findings, requiring domain knowledge and result interpretation beyond code quality.

MemCoE: cognition-inspired two-stage memory optimization for LLM agents to learn personalized long-term user preferences within context windows.

Analysis of agentic AI specialization: coding agents (Codex-style) for knowledge work, Claude for creative tasks; discusses agents escaping operational boundaries.

Intern-Atlas introduces structured methodological evolution graphs as research infrastructure for AI agents to navigate scientific knowledge beyond citation links.

Link lets users connect cards, banks, and subscriptions, then authorize AI agents to spend securely via approval flows.

STEF enables schema-agnostic evaluation of text-to-SQL agents in production without ground-truth queries, addressing real-world deployment gaps.

Study finds persona prompting in multimodal LLMs produces stable but limited behavioral variation in urban sentiment judgment tasks.

CARE methodology systematizes LLM agent engineering in scientific domains via three-party collaboration between SMEs, developers, and helper agents.

NVIDIA CUDA Tile (cuTile) is a tile-based programming model that enables developers to write GPU kernels in terms of tile-level operations—loads, stores, and... NVIDIA CUDA Tile (cuTile) is a tile-based programming model that enables developers to write GPU kernels in terms of tile-level operations—loads, stores, and matrix multiply-accumulate—rather than manually coordinating threads, warps, and shared memory. cuTile.jl brings the same tile-based approach to the dynamic programming language Julia. Users can write custom GPU kernels without dropping… Source

Architectural pattern language for vision language agents balances latency/non-determinism of VLMs against real-time enterprise control requirements.

Comparative evaluation of three LLM agent paradigms (domain-specific, computer-use, coding) on scientific visualization tasks across 15 benchmarks.

D3-Gym dataset: 565 verifiable tasks from real scientific repositories for evaluating LLM agents on data-driven discovery.

Language model agents design mechanical linkages via symbolic lifting: LLMs explore topologies while numerical optimizers tune parameters, validated on six motion targets.

Survey of RL+GUI agents for long-horizon automation in visual interfaces; proposes framework toward autonomous digital inhabitants with safe exploration.

Schema-grounded external memory for agents outperforms text-retrieval approaches by enabling exact fact tracking, state updates, and structured queries.

Survey formalizing graph-based world models for agents, decomposing environments into entity nodes and edges to improve robustness vs. flat-tensor approaches.

On-demand persona-based agent generation framework enabling dynamic multi-agent workflow customization without hard-coded architectures.

Lightweight clinical agent architecture using integrated state dynamics to surface pre-escalation risk signals in LLM clinical deployment.

KellyBench: long-horizon sequential decision benchmark using 2023-24 Premier League sports betting; evaluates agents on non-stationary open-ended optimization.

Working on large codebases with Claude Code, we kept running into the same issue: when Claude looks for relevant code, it falls back to grep, reading full files, or launching multiple subagents. This burns through tokens, and often misses the relevant code. There are some existing solutions (that we also benchmarked against), but they all had issues (too slow, needs API keys, quality not good enough, etc). We built [Semble](https://github.com/MinishLab/semble) to fix this. It's a local MCP server that gives Claude Code high quality code search: instead of reading files to find what's relevan...

tl;dr: your skill in AI is a measure of your **quality** and **scale**. Use **success criteria** and **subagents** intentionally to get excellent results. Use skills and .md docs when you find repeating patterns in your daily work, not before. **---** **Quality** comes from telling the agent what outcome you want, and the **success criteria** that you will use to measure a “good” outcome. This helps avoid Claude's tendency to rush completion. Note this is specifically *not* telling it what to *do*, but instead what to *achieve*. If you come from the old world, you might remember terms like ...

I can have multiple, dense legal documents on my screen, each 40, 60, or 100+ pages each with the Claude Word add-in agents syncing, pushing and pulling information between them, pinging each other, and providing helpful context so that I can draft all three or four in parallel or ensure that an entire package is consistent. I can have a lengthy spreadsheet workbook open containing 10 worksheets and the information is analyzed and pulled in by the agents when needed. I am absolutely blown away at how well this is implemented and the improvement in quality, consistency and efficiency. It ...

ClawGym framework for systematic development of file/tool-based agents, includes 13.5K synthesized tasks for agent training and evaluation.

Operating and maintaining (O&M) large-scale online engine systems (search, recommendation, advertising) demands substantial human effort for release monitoring, alert response, and root cause analysis. While LLM-based agents are a natural fit for these tasks, the deployment bottleneck is not reasoning capability but orchestration: selecting, for each operational event, the relevant data (metrics, logs, change events) and the applicable operational knowledge (handbook rules and practitioner experience). Feeding all signals indiscriminately causes dilution and hallucination, while manually cura...

FutureWorld: live environment for training LLM-based agents on real-world outcome prediction and continual learning.

Mistral AI launches Mistral Medium 3.5 with remote coding agents in Vibe and Work mode in Le Chat for complex tasks.

We visited Scout AI's training ground where it's working on AI agents that give individual soldiers control of fleets of autonomous vehicles.

DV-World benchmark evaluates data visualization agents across 260 real-world tasks spanning spreadsheet manipulation, chart creation, and dashboard repair.

Sam Altman and Matt Garman discuss OpenAI-AWS partnership on Bedrock Managed Agents; Stratechery covers OpenAI-Microsoft deal implications.

ADEMA architecture enables long-horizon LLM-agent tasks via explicit knowledge-state bookkeeping, dual-evaluator governance, and checkpoint-resumable persistence.

Agora-Opt combines decentralized multi-agent debate with memory-augmented LLMs for automated optimization modeling from natural-language requirements.

SkillSynth automates terminal task synthesis via skill graphs to improve trajectory diversity for training command-line execution agents.

Salesforce production inference architecture for compound AI systems supporting heterogeneous model composition, agents, and retrieval at scale.

CORAL framework integrates neurocognitive governance principles into autonomous AI agents for safety-critical deployment with internalized behavioral alignment.

NeLLCom-Lex framework models human color naming lexicons in neural agents; extends with context modeling to reduce non-convex divergence from human categories.

Tank OS puts OpenClaw AI agents into a container that let's it run reliably and more safely, especially for those running fleets of them.

SnapGuard detects prompt injection attacks on screenshot-based web agents using lightweight multimodal methods instead of large VLMs.

Semantic Gateway framework applies formal validation and zero-trust security to LLM-orchestrated enterprise APIs using Model Context Protocol.

Automated adversarial collaboration framework using LLM agents and program synthesis to adjudicate competing cognitive science theories.

OpenAI GPT models, Codex, and Managed Agents now available on AWS for enterprise deployment.

Persona Collapse in multi-agent LLM simulations: agents converge to homogeneous behavior despite distinct profiles; framework measures Coverage, Uniformity, Complexity.

SciCrafter: Minecraft benchmark evaluating agents' discovery-to-application loop via parameterized redstone circuit tasks.

Informational Viability Principle for autonomous AI agent governance: runtime monitoring and restriction via unobserved risk bounds without code changes.

AgentWard: defense-in-depth lifecycle security architecture for autonomous AI agents spanning initialization through execution.

Skill Retrieval Augmentation enables LLM agents to retrieve relevant skills from large corpora without explicit enumeration.

QA engineer discusses challenges testing non-deterministic LLM agents in production, seeking rigorous evaluation methods beyond traditional assertion-based testing.

China has ordered Meta to unwind its multibillion-dollar Manus acquisition, dealing a potential setback to Zuckerberg’s push into AI agents.

The phone could go in mass production in 2028, an analyst says.

Google and Kaggle launch 5-day AI Agents Intensive Course; registration open.

Choco uses OpenAI APIs to automate food distribution logistics via AI agents, improving productivity.

Reddit user seeks advice on setting up local coding agents like Claude Code with open-weight models via llama.cpp.

Recent evidence suggests that frontier AI systems can exhibit agentic misalignment, generating and executing harmful actions derived from internally constructed goals, even without explicit user requests. Existing mitigation methods, such as Reinforcement Learning from Human Feedback (RLHF) and constitutional prompting, operate primarily at the model level and provide only probabilistic safety guarantees. We propose the Policy-Execution-Authorization (PEA) architecture, a "separation-of-powers" design that enforces safety at the system level. PEA decouples intent generation, authorization, an...

In a recent experiment, Anthropic created a classified marketplace where AI agents represented both buyers and sellers, striking real deals for real goods and real money.

[http://claude.ldlework.com](http://claude.ldlework.com/) I built this for myself but I figured why not share. I'm happy to receive feedback, I know it's not perfect. Thanks for taking a look. The aim of CCM is to be able to fully manage all Claude Code configuration files, both globally and those in your project. Some neat features: \- Manages your [CLAUDE.md](http://claude.md/), rules, hooks, agents, memories and so on. \- Elevate memories to rules \- Copy/Move any asset from one scope to another, or elevate it to global scope \- Install marketplaces and plugins The full app is embe...

Systematic analysis of token consumption patterns in agentic coding tasks across eight frontier LLMs on SWE-bench Verified.

Taxonomy of world modeling capabilities for AI agents across three levels (predictor, simulator, reasoner) organized by environmental laws.

SOLAR-RL bridges offline and online RL for training MLLM GUI agents on dynamic tasks, combining trajectory semantics with long-horizon learning.

Agents are amazing. Harnesses are cool. But the fundamental role of a data scientist is not to use a generalist model in an existing workflow; it's a completely different field. AI engineering is the body of the vehicle, whereas the actual brain/engine behind it is the data scientist's playground. I feel like I am not alone in this realisation that my role somehow got silently morphed into that of an AI engineer, with the engine's development becoming a complete afterthought. Based on industry requirements and ongoing research, most of the work has quietly shifted from building the engine t...

Chinese AI company DeepSeek released a preview of its hotly anticipated next-generation AI model V4 on Friday, saying that the open-source model can compete with leading closed-source systems from US rivals including Anthropic, Google, and OpenAI. DeepSeek says V4 marks a major improvement over prior models, especially in coding, a capability that has become central to AI agents and helped drive the success of tools like ChatGPT Codex and Claude Code. The release is also a milestone for China's chip industry, with DeepSeek explicitly highlighting compatibility with domestic Huawei technology....

In March 2026, three LLM agents generated over 600,000 lines of code, ran 850 experiments, and helped secure a first-place finish in a Kaggle playground... In March 2026, three LLM agents generated over 600,000 lines of code, ran 850 experiments, and helped secure a first-place finish in a Kaggle playground competition. Success in modern machine learning competitions is increasingly defined by how quickly you can generate, test, and iterate on ideas. LLM agents, combined with GPU acceleration, dramatically compress this loop. Historically… Source

Nemobot is an interactive environment for creating and deploying LLM-powered game agents across multiple game classes using Claude Shannon's taxonomy.

StructMem proposes hierarchical memory framework for LLM agents balancing relational structure preservation with efficiency for long-horizon reasoning.

Agent Evolving Learning framework enables LLM agents to accumulate and leverage experience across open-ended episodes via two-timescale Thompson sampling.

Earlier this month, millions of OpenClaw users woke up to a sweeping mandate: The viral AI agent tool, which this year took the worldwide tech industry by storm, had been severely restricted by Anthropic. Anthropic, like other leading AI labs, was under immense pressure to lessen the strain on its systems and start turning a profit. So if the users wanted its Claude AI to power their popular agents, they'd have to start paying handsomely for the privilege. "Our subscriptions weren't built for the usage patterns of these third-party tools," wrote Boris Cherny, head of Claude Code, on X. "We wa...

N-gram models match LSTM/Transformer accuracy on event-log prediction with lower resources and better stability than neural baselines.

I spent some time few days back comparing Opus 4.6 and 4.7 using my own usage data - just to see how they actually behave side by side. [https://github.com/getagentseal/codeburn](https://github.com/getagentseal/codeburn) it’s still pretty early for 4.7, but a few things surprised me. In my sessions, 4.7 gets things right on the first try less often than 4.6. One-shot rate sits around 74.5% vs 83.8%, and I’m seeing roughly double the retries per edit (0.46 vs 0.22). It also produces a lot more output per call - about 800 tokens vs 372 on 4.6 - which makes it noticeably more expensive. ...

Open-source Agent-Quest tool visualizes Claude Code agents in real-time across parallel CLI sessions using fantasy-themed 2D UI.

OpenAI is giving users of its Business, Enterprise, Edu, and Teachers plans access to cloud-based "workspace" agents available in ChatGPT that can perform business tasks. In its blog post, OpenAI gives examples of agents like one that finds product feedback on the web and sends a report in Slack and a sales agent that can draft follow-up emails in Gmail. These new agents follow increasing interest in agents across the AI landscape, especially after OpenClaw - the AI agent formerly known as Clawdbot and Moltbot that touts itself as the "AI that actually does things" - went viral. OpenClaw foun...

Interval POMDP shielding method for autonomous systems with learned perception, using confidence intervals to block unsafe actions under sensor uncertainty.

Meta employees' activity at work is now being used to train the company's AI agents. As reported by Reuters, Meta is installing a tool it calls Model Capability Initiative (MCI) on US-based employees' computers that runs in work-related apps and websites, recording mouse movements, clicks, keystrokes, and occasional screenshots. The data from this tool will be used to train the company's AI models to get better at interacting with computers the way humans do, including automating work tasks like those Meta's employees perform on the job. According to Reuters, the data from MCI won't be "used ...

LLM agents in repeated Avalon deception games develop reputation dynamics and social memory across 188 games, studying emergent multi-round behavior.

ProactAgent enables lifelong learning agents to proactively retrieve past experience and skills during task interaction rather than passively.