Tool use, function calling, orchestration and the protocols that let models act rather than only answer.
Medical artificial intelligence agents should retrieve patient identifiers only when a task requires them, the data-minimization principle behind the HIPAA minimum necessary standard for protected hea...
The article discusses how modern architectures create precise, cost‑efficient context for enterprise AI agents, moving beyond Retrieval‑Augmented Generation (RAG). It presents six advanced architectural patterns that serve as a practical guide for practitioners. The focus is on engineering context layers that improve the performance and efficiency of AI systems in enterprise settings.
The release of datasette-llm 0.1a9 introduces a small user‑experience improvement: the plugin now automatically detects the default LLM model, eliminating the need for users to manually configure a default model when running Datasette Agent. This change simplifies the setup process for users who previously had to specify a model with the `-s plugins.datasette-agent.default_model` option.
The article discusses a research paper suggesting that reward labels have minimal impact on how AI agents learn. The author tested this claim on three different models, finding that one model behaved exactly as the incorrect reward labels instructed. The piece highlights how a single model can be misled by flawed feedback.
DecompGym is a benchmark that tests large language model agents on designing Benders decomposition and column generation algorithms. The benchmark contains 81 instances across nine problem families, requiring agents to produce solvers that run within 60 seconds, pass a method audit by an LLM judge, and achieve an objective within 1% of the optimal value. Evaluation of 12 GPT, Claude, and Gemini models shows that while some models solve all instances on the first try, many fail due to missing or suboptimal solutions, and reflection-based debugging improves success rates compared to simple regeneration.
CALM is a conflict‑aware lifecycle memory governance framework designed for long‑horizon language agents. It transforms heterogeneous records into evidence‑grounded memory cells, updates their validity and lifecycle states using support, conflict, and dependency signals, and admits query‑appropriate evidence through training‑free or learned gating based on temporal applicability. Experiments on LongMemEval‑S and LoCoMo with Qwen and GPT backbones show that CALM outperforms state‑of‑the‑art baselines, improves evidence selection, suppresses erroneous memories, and preserves superseded records for accurate earlier‑state reasoning.
The paper proposes the Explicit Belief Multi-Agent System (EB‑MAS) to address safety in large language model (LLM) powered multi‑agent systems. It treats belief as an engineering abstraction for evidence‑based commitments and externalizes belief states into a structured substrate, using belief operators to propose updates and a Cognitive Firewall to filter weakly evidenced changes. The authors also highlight risks such as manufactured consensus, collusion, and cascading misinformation, and discuss practical challenges in grounding, false positives, scalability, and ethical misuse.
The paper proposes a three‑engine architecture for reliable financial AI, combining a financial knowledge graph, a deterministic optimization engine, and a multi‑agent LLM layer. A coordination layer called context engineering selects and validates query‑specific context from ERP systems and other tools. Observations from 15 deployments show that multi‑agent validation reduces hallucination rates to below 1%, while reusable mappings and templates cut deployment time from six to two weeks.
EpiAgent introduces an explicit, learnable epistemic layer for large language model agents, enabling them to maintain and update beliefs about other agents’ private information. The framework splits the policy into Epistemic Update and Action Planning stages, optimized via a bi‑level Epistemic Self‑Play algorithm that rewards accurate belief predictions. Experiments in mixed and competitive settings show EpiAgent outperforms prompting and RL baselines, highlighting the benefits of dedicated epistemic training for belief inference and action planning.
The article discusses how AI agents are transforming software development by autonomously creating pull requests, selecting dependencies, refactoring code, interpreting test failures, and managing infrastructure across the SDLC. It highlights the lack of accountability for these transient processes, noting that traditional tools like git logs and governance frameworks assume human decision‑making, which no longer applies. To address this, the authors propose Agentic Process Evidence (APE), a signed, tamper‑evident, machine‑readable attestation that links AI agent sessions to their outcomes, enabling policy‑as‑code controls and human oversight only when necessary.
DynaRev is a new benchmark comprising 181 tasks designed to evaluate web agents when users revise their goals during interaction. It introduces semantic milestones to trigger additive, unconditional, and runtime-conditioned revisions, and assesses agents on correctness, responsiveness, and efficiency via trajectory‑grounded evaluation. Experiments with five models show that agents often only partially satisfy revised tasks, with performance varying by task source, revision type, and model, and highlight the benefits of instruction rewriting versus browser resets.
OpenFrontiers is a cross‑disciplinary resource that compiles 12,151 open research problems and a benchmark of 200 recently resolved mathematics problems, including 110 solved in 2026 and 90 still open. Each benchmark problem is formally stated in Lean 4, verified by automated probes, and reviewed independently. The evaluation protocol tests LLM agents under closed‑book, frozen‑retrieval, and live‑web conditions, revealing that closed‑book rediscovery is rare while live‑web success largely reflects retrieval of published solutions.
The paper investigates how different types of ontologies—curated, LLM-generated, and data-derived—serve as checkers for large language model agents answering questions over knowledge graphs. By evaluating these ontologies as prompt context, as checkers, or both, across both open-weight and proprietary agents on public and a large private graph, the study quantifies the trade-offs between errors caught and false alarms. It finds that curated ontologies can flag more correct answers than other sources but are only marginally more precise than random, and that combining sources or providing specific feedback offers limited benefits.
The paper introduces Partner‑Outsider Positive Listening (PoPLi), a behavioural test that detects covert collusion between AI agents by comparing how a partner agent and an uninvolved outsider react to reworded messages. PoPLi measures the difference in action changes when the message’s wording is altered while its meaning remains the same, and it is calculated post‑hoc from logged prompts. In simulations, PoPLi reliably distinguishes scenarios with a covert code from those without, whereas message‑only monitors fail to do so.
Seq@K is a new evaluation protocol that measures a language model’s ability to solve a problem within K sequential attempts, receiving feedback after each failure. It isolates the core capability of iterative refinement from feedback, contrasting with existing metrics like Pass@K that do not capture this behavior. Experiments on seven benchmarks and sixteen frontier models show that Seq@K can change model rankings, reveal more efficient configurations, and provide insights into how models use feedback.
The paper introduces GARDEN, a framework that uses large language models to autonomously design a single reward program for cooperative multi‑agent reinforcement learning tasks. By navigating evidence from environment code, training logs, and agent behavior, the reward is tested and finalized without altering the learning algorithm. Experiments on Overcooked and SMAX benchmarks show that the autonomously engineered reward improves task returns by up to 58.2% and win rates by up to 89.9% compared to default rewards, outperforming prior methods.
SkillMem introduces a compact external latent memory for reusable skills in latent multi‑agent systems, enabling role‑conditioned cross‑attention to guide agents without storing large per‑skill parameters. Trained via behavioral distillation, SkillMem matches the effect of textual skills while keeping the backbone frozen. Across six benchmarks and Qwen3 models from 4B to 14B, it delivers strong gains for smaller backbones, reduces skill storage by over 38×, and achieves up to 4.23× inference speedup.
The study demonstrates that multi‑turn reinforcement learning (RL) in simulation environments significantly improves a language model’s performance on structured, text‑based psychiatric intake tasks. In a randomized, double‑blind crossover trial involving 60 matched scenario pairs, the RL‑trained agent outperformed its base model in 76.7% of cases, showing higher scores on clinical examination, patient‑centered communication, and differential diagnosis appropriateness. Across 127 validated simulated scenarios, the trained agent nearly doubled the base model’s clinical rubric score and reduced under‑triage of acute safety risks by more than half, indicating that simulation provides a controlled testbed for training and stress‑testing psychiatric conversational agents.
The paper introduces ClinGeX, a large‑language‑model agent that uses 105 curated executable skills to automate clinical genomic analysis. In benchmark tests, ClinGeX achieved 71.7% macro‑averaged task‑level concordance, outperforming the best comparator at 45.7%. Human‑AI collaboration studies showed ClinGeX significantly improved case‑level concordance across variant analysis, rare disease interpretation, and oncology decision‑making, while also reducing interpretation time.
VeriCUA is a platform-agnostic framework that treats verification of computer-using agents as dynamic evidence gathering and assessment. It constructs a task rubric from available context, then selects GUI or code checks to resolve unmet requirements and combines their findings under explicit evidence rules. Experiments demonstrate that VeriCUA achieves higher agreement with reference labels than baseline methods, especially on hard offline cases and in rejecting false completion claims online.