547 episodes, newest first. Each one covers a single research paper.
- EP548: How AI Drones Search Using Spatial Imagination
Nov 12, 2026 · SearchWorld: Spatial Value-Grounded Imagination for UAV Object Search via World Models
- EP547: Why AI Confuses Brainstorms With Facts
Nov 12, 2026 · Whose Memory Is It? Scope-Aware Commit Rules for Long-Term LLM Memory
- EP546: How Self Driving AI Grades Itself
Nov 11, 2026 · VeriFine: Scaling Verification for Self-Improvement in Embodied Reasoning
- EP545: S1-MAS slashes multi-agent AI token waste
Nov 11, 2026 · Token-Efficient Multi-Agent Collaboration via System One-Guided Computational Division of Labor
- EP544: DEORCH Decouples Multi-Agent AI Teams
Nov 10, 2026 · Decoupled Multi-Agent Orchestration
- EP543: Fixing AI Memory for Unseen Environments
Nov 10, 2026 · MemCo: Memory-Centric Collaboration for Generalizing LLM Agents to Unseen Environments
- EP542: LightMTP Teaches AI to Plan Ahead
Nov 9, 2026 · LightMTP: Lightweight Latent Multi-Token Prediction
- EP541: How HEAR Protocol Speeds Up AI Agents
Nov 9, 2026 · Can Agent Harnesses and Inference Engines Hear Each Other? The HEAR Protocol for Agentic LLM Serving
- EP540: How AI Robots Dream Before Moving
Nov 8, 2026 · PreAct-Nav: Agentic Reasoning Before Action for Urban Navigation
- EP539: How GitSwarm Cures AI Amnesia
Nov 8, 2026 · GitSwarm: Decentralized Compounding Inference
- EP538: How DAGO Slashes AI Workflow Costs
Nov 7, 2026 · Fusion is the New Mutation: Bandit-Guided Evolution on Workflow Graphs
- EP537: AI Rewrote Its Engine to Cheat Benchmarks
Nov 7, 2026 · SEIS: Self-Evolving Inference Systems
- EP536: Why AI Benchmarks Hide Agent Failures
Nov 6, 2026 · Agentic Cognitive Depth: Operational Criteria for Evaluating LLM Agents
- EP535: AI Counselors That Evolve in Real Time
Nov 6, 2026 · PsyEvo: A Personalized Counseling Agent That Self-Evolves at Test Time
- EP534: Robots That Learn to Fix Their Mistakes
Nov 5, 2026 · MobiAgent: Dual-Loop Recursive Policy Self-Improvement for Long-Horizon Mobile Manipulation
- EP533: Why Smarter AI Invents Alien Languages
Nov 5, 2026 · On Language Drift during RLVR Post-Training
- EP532: MemFit Reinvents AI Agent Memory
Nov 4, 2026 · MemFit: Efficient Long-Term Agentic Memory
- EP531: Fixing multi-step AI without expensive critics
Nov 4, 2026 · T2SPO: Trajectory-to-Step Policy Optimization for Agentic Reinforcement Learning
- EP530: JevSpawn Solves the AI Token Bottleneck
Nov 3, 2026 · JevSpawn: Adaptive Agentic Inference through Compositional Action Spaces
- EP529: Single AI Agents Beat Complex Frameworks
Nov 3, 2026 · How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering?
- EP528: How Hindsight Makes Small AI Beat GPT
Nov 2, 2026 · Beyond Prediction: Steering VLM Agents with Retrospective World Modeling
- EP527: How Confidence Filtering Fixes AI Reasoning
Nov 2, 2026 · Provable Test-Time Scaling for Beam Search in LLM Reasoning
- EP526: From Model Distillation to Agent Distillation
Nov 1, 2026 · Distilling Agentic Systems: A Roadmap across Models, Artifacts, and Harnesses
- EP525: How AnyAct Fixes Desktop AI Agents
Nov 1, 2026 · AnyAct: Universal Action for Self-Evolving Agents
- EP524: How SkillRubric Helps AI Grade Itself
Oct 31, 2026 · SkillRubric: Co-Evolving Actor Guidance and Evaluator Rubrics for Multimodal Agents
- EP523: How Subgoals Stop AI Belief Drift
Oct 31, 2026 · SGG-ReflAct: Sub-Goal Guided ReflAct with Structured Planning for Reliable Long-Horizon Reasoning
- EP522: Why Limited Memory Makes AI Smarter
Oct 30, 2026 · Continuous Context Management
- EP521: HTN Planning Fixes AI Tool Orchestration
Oct 30, 2026 · HTN Planning as a Coordination Layer for Multi-Server MCP Tool Orchestration
- EP520: How CodeSkill Solves Long Horizon Coding
Oct 29, 2026 · CodeSkill: Latent Skill Abstraction for Long-Horizon Code Agents
- EP519: Multi Agent AI for Financial Valuation
Oct 29, 2026 · Self-Evolving Multi-Agent Symbolic Discovery for Financial Fundamental Analysis
- EP518: Teaching Robots Without Human Puppeteers
Oct 28, 2026 · CAPEX: Efficiently Distilling Foundation Model Behavior into Deployable Robot Policies through Experience-Adaptive Reasoning
- EP517: How DynBranch Eliminates AI Wait Times
Oct 28, 2026 · DynBranch: Speculative Subgraph Reuse for Dynamic Agentic LLM Serving
- EP516: ActKV Solves AI Agent Memory Overload
Oct 27, 2026 · ActKV: Efficient LLM Agents through Action-Guided KV Cache Management
- EP515: How IterSynth Solves AI Context Exhaustion
Oct 27, 2026 · IterSynth: Rethinking Deep Search Agents via Role-Decoupled Iterative Synthesis
- EP514: How AI Agents Fix Their Own Mistakes
Oct 26, 2026 · A Wrong Turn Does Not Ruin the Journey: Deviation-Guided Skill Self-Evolution for LLM Agents
- EP513: How AI Drone Swarms Fly Without Text
Oct 26, 2026 · Controlling Collectives of AI Agents in Reasoning Space with Spatial Transformers
- EP512: How Agent Editing Fixes AI Reasoning
Oct 25, 2026 · Agent-Editing World Model: Rethinking World Modeling for LLM Agents
- EP511: How Human Explanations Teach AI Causality
Oct 25, 2026 · xWhyL: Causal Interactive Learning
- EP510: When AI Can Trust Its World Model
Oct 24, 2026 · Dual-Frontier: When Can an Agent Trust Its World Model?
- EP509: How AI Video Targets Cognitive Vulnerabilities
Oct 24, 2026 · CogenPVG: Cognitive-Enhanced Reflective Multi-Agent Framework for Persuasive Video Generation
- EP508: Training Small Models to Steer LLM Reasoning
Oct 23, 2026 · Beyond Repeated Sampling: Learning Search Policies for LLM Reasoning
- EP507: How X-Planner Makes Laundry Robots Real
Oct 23, 2026 · X-Planner: Event-Structured Task Planning for Embodied Intelligence
- EP506: How AI Redesigns Its Own Harness
Oct 22, 2026 · RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
- EP505: How Data Agents Make Databases Autonomous
Oct 22, 2026 · Data Agents: Agentic Data Systems
- EP504: How RPMem Cures AI Amnesia
Oct 21, 2026 · RPMem: Learning Long-Term Recurrent Parametric Memory Across Sessions for LLM Agents
- EP503: How FLARE Fixes AI Coding Errors
Oct 21, 2026 · FLARE: A Full-Lifecycle Dense Supervision Paradigm for Long-Horizon Coding Agents via Generative Reward Model
- EP502: Sub Second AI Compression Without Training Data
Oct 20, 2026 · Neural Spectral Capacity: Measuring and Designing Architectures from Network Specification Alone
- EP501: How Lazy Execution Cuts AI Agent Costs
Oct 20, 2026 · LazyAgent: Demand-Driven Materialization and Physical Optimization of Agentic Programs
- EP500: The Hidden World Map Inside TaxiGPT
Oct 19, 2026 · World Modeling in Transformers
- EP499: CodeMidas Trains AI Agents From Raw Code
Oct 19, 2026 · CodeMidas: Scaling Agentic Coding RL Environments from Code Itself
- EP498: How AI Travel Agents Repair Disrupted Itineraries
Oct 18, 2026 · Replan, Repair, or Edit? A Unified Empirical Evaluation of Travel Agents for Itinerary Revision under Resource Disruptions
- EP497: Why AI Agents Lie About Reading Files
Oct 18, 2026 · Quantifying Overclaiming Propensity in Frontier LLM Agents
- EP496: How LLM Wiki Fixes AI Memory
Oct 17, 2026 · WFM: Wiki Foundation Model for Complex Agentic Reasoning
- EP495: How SynAgent Runs Physical Lab Experiments
Oct 17, 2026 · Hypothesis-Driven Autonomous Materials Synthesis with Multimodal LLM Agents
- EP494: The Ten Thousand Dollar AI Test
Oct 16, 2026 · Beyond Outcomes: Dual-View Relational Learning for Efficient Agent Benchmarking
- EP493: How AI Rewires Its Own Personality
Oct 16, 2026 · Self-Emergence Agent Architecture:Behavior-Inertia HMM, Reflexive Metacognition,and Social-Contrastive Self-Modeling
- EP492: EchoPath Gives AI Digital Muscle Memory
Oct 15, 2026 · EchoPath: Execution-Level Replayable Memory for GUI Agents
- EP491: Why AI Agents Need a Social Harness
Oct 15, 2026 · Agentic Societies Need a Social Harness
- EP490: How VideoScout Watches Long Videos
Oct 14, 2026 · VideoScout: Learning Agentic Active Exploration with Adaptive Reasoning Pacing for Long Video Understanding
- EP489: How AlgoEvo Builds Self-Evolving AI Algorithms
Oct 14, 2026 · AlgoEvo: Self-Evolving Agentic Search for Automated Algorithm Discovery
- EP488: How ANASSA Fixes Geospatial AI
Oct 13, 2026 · ANASSA: An Agentic AI Orchestration Framework for Spatial Intelligence
- EP487: How AI Memory Rewires Like Human Brains
Oct 13, 2026 · Retrieval-Driven Memory Reconsolidation for Long-Term LLM Agents
- EP486: How ECDYSIS Fixes Self-Evolving AI Agents
Oct 12, 2026 · Ecdysis: Efficient and Effective Training of Runtime Harnesses for LLM Agents
- EP485: How ConvMem Fixes AI Goldfish Memory
Oct 12, 2026 · ConvMem: Convolutional Memory for Long-Context Reasoning
- EP484: Fixing LLM Planning with Math Engines
Oct 11, 2026 · Belief-State Engine: Augmenting LLMs for Principled Planning Under Partial Observability
- EP483: How Procedural Graphs Fix AI Amnesia
Oct 11, 2026 · Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
- EP482: Training AI with Feedback Enriched Environments
Oct 10, 2026 · Environments as Scaffold: Enriching Feedback to Bootstrap Self-Evolving Agents in Long-Horizon Tasks
- EP481: Why Expert Imitation Breaks Small AI Models
Oct 10, 2026 · Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails
- EP480: Cloning Powerful AI Agents Just by Watching
Oct 9, 2026 · AgentLeak: Cloning Stronger LLM Agent Capabilities onto Weaker Agents Beyond Skill Stealing
- EP479: Multi-Agent AI Debates Are Pure Theater
Oct 9, 2026 · A Layered Analysis of Disagreement And Answer Quality in Multi-Agent LLM Debate
- EP478: How Geometry Fixes AI Memory Limits
Oct 8, 2026 · ECOKV: Geometry-Aware KV Cache Eviction via Complementary Diversity Metrics
- EP477: Why AI Decision Explanations Are Broken
Oct 8, 2026 · Causal Attribution for Agentic Decisions: Estimators, Coupling, and a Traceability Specification
- EP476: Hardcoding Values into AI Architecture
Oct 7, 2026 · Value-Preserving Architectures for Agentic AI Systems
- EP475: Von Neumann Automata for Self Reproducing AI
Oct 7, 2026 · Dalek: A Constructive Agent Machine
- EP474: How SkillGLoW Fixes AI Memory Hoarding
Oct 6, 2026 · SkillGLoW: Procedural-Family Skill Consolidation for Self-Improving Agents on Long-Horizon Task Streams
- EP473: How six templates make AI 120x faster
Oct 6, 2026 · Codebook Agent: Amortized Topology Design for LLM Multi-Agent Systems
- EP472: Why AI Reasons Better in Silence
Oct 5, 2026 · Latent Recurrent Thoughts: Recurrent Refinement of Proposed Latents for Reasoning with Frozen LLMs
- EP471: Self evolving autonomous agents with ARISE-RL
Oct 5, 2026 · ARISE-RL: Agentic Rubric-Grounded Iterative Self-Evolution with Reinforcement Learning
- EP470: 1933 math fixes AI context limits
Oct 4, 2026 · Higher-Dimensional Rotary Position Embedding
- EP469: How AutoCRAT stops AI from overthinking
Oct 4, 2026 · AutoCRAT: Within-trajectory Joint Control of Stochasticity and Compute for LLM Reasoning
- EP468: Steering AI logic with latent vectors
Oct 3, 2026 · Toward Latent Language Model Skills Steering and Optimization: An Empirical Study
- EP467: LiteSearch-VL gives small AI search powers
Oct 3, 2026 · LiteSearch-VL: Small Multimodal Search Agents via Trajectory Distillation and Synthetic Step-DPO
- EP466: Sony OmniUE replaces typing with physical interaction
Oct 2, 2026 · Omni-Interactive Universal Embedder
- EP465: AI agents finally ditch computer screens
Oct 2, 2026 · ASIL: Replacing Screenshot-and-Click with Structured State and Semantic Actions
- EP464: PROGROUTER manages the cost of AI agents
Oct 1, 2026 · ProgRouter: Online Progress-Guided Orchestration for Multi-Agent LLM Workflows under Quality-Cost Tradeoffs
- EP463: How PolyMemDB Solves AI Memory Hallucinations
Oct 1, 2026 · PolyMemDB: A Polyglot Database System for AI Memory Management
- EP462: OpsHarness turns general AI into SRE experts
Sep 30, 2026 · From General Agents to RCA Experts: A Self-Evolving Harness for Root Cause Analysis
- EP461: AsymSpec Slashes AI Agent Compute Costs
Sep 30, 2026 · AsymSpec: Context-Asymmetric Speculative Decoding for Agentic LLMs
- EP460: Teaching AI to Ignore Its Teacher
Sep 29, 2026 · On-policy Distillation with Verifiable Reward
- EP459: Small models outsmart giants by coding
Sep 29, 2026 · Joint Optimization of Tool Creation and Use for Large Language Model Agents
- EP458: Building AI agents that know their limits
Sep 28, 2026 · TRACE: A Self-Evolving Skill Bank for Consistent, Limit-Aware LLM Agents
- EP457: How AI finally masters 3D space
Sep 28, 2026 · Object-Uni: A Unified Model for Object-Centric Spatial Understanding and Controllable Generation
- EP456: Why AI robot crews need a bouncer
Sep 27, 2026 · Physical Agentic AI: An Architecture for Orchestrating a Robot Crew with LLMs
- EP455: Stopping Coalition Pollution In AI Agents
Sep 27, 2026 · Coalition-Aware Skill Reliability for Self-Evolving Agents
- EP454: AI is eating the software stack
Sep 26, 2026 · The Third Restructuring of Software Form: From the Three-Tier Architecture to Storage, Models, and Agents
- EP453: PolicyGuide forces AI agents to follow workflows
Sep 26, 2026 · PolicyGuide: From Guarding One Action to Guiding the Whole Workflow for Policy-Compliant LLM Agents
- EP452: A Budget for Smarter AI Reasoning
Sep 25, 2026 · Training-Free Inference-Time Self-Reflection and Cost-Bounded Early Stopping for Large Language Models
- EP451: Breaking the AI environment bottleneck with SPADE
Sep 25, 2026 · SPADE: Self-Play in Adaptive Synthetic Executable Environments
- EP450: AI models grade each other to scale
Sep 24, 2026 · Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL
- EP449: Why AI Physics Code Hits Invisible Walls
Sep 24, 2026 · An Omitted Mode Is a Rare Rule: The Sampling-Verification Danger Law in Continuous Code World Models
- EP448: TDD-Agent and test-driven AI reasoning
Sep 23, 2026 · TDD-Agent: Test-Driven Reasoning for Code Generation
- EP447: Building muscle memory for AI agents
Sep 23, 2026 · HaReCAP: Habitual-action Grounding for Recursive Large Language Model Agents
- EP446: Agent-Native Telemetry for Autonomous Cloud Operations
Sep 22, 2026 · Agent-Native Telemetry: Verifiable State-Delta Evidence for Autonomous Operations
- EP445: How HyMem stops AI context dilution
Sep 22, 2026 · HyMem: Hierarchical Context Management for Long-Horizon Agents via Information Isolation
- EP444: EgoGazeLite replaces bulky eye tracking hardware
Sep 21, 2026 · EgoGazeLite: On-Device Egocentric Gaze Prediction for Token-Efficient Multimodal LLM Video Input
- EP443: Securing Autonomous AI Through Infrastructure
Sep 21, 2026 · Bounded Agents: Delegation Security for Multi-Agent AI Systems
- EP442: Mobius Separates AI Knowledge and Reasoning
Sep 20, 2026 · Intern-S2-Mobius: Foundation Model with Decoupled Knowledge and Reasoning
- EP441: Making AI Agents Reliable with ACID
Sep 20, 2026 · Agentic Transaction: Towards ACID-Compliant Agent Systems
- EP440: Direct latent communication between AI agents
Sep 19, 2026 · StateBridge: Training-free Hidden-state Alignment for Latent Communication in LLM Multi-Agent Systems
- EP439: AI internal math distinguishes lies from impossibilities
Sep 19, 2026 · Falsehood and Impossibility Are Different Directions in an AI's Representation of Language
- EP438: Ready Cohorts Stop AI Agent Stutter
Sep 18, 2026 · Ready Cohorts: Bounding GPU Opportunity and Avoiding Host Round Trips in LLM-Agent Control
- EP437: Giving AI agents an undo button
Sep 18, 2026 · LoongReflect: Boosting Long-Horizon Reflection in Search Agents via Global Perspective Distillation
- EP436: Giving drones directions in 3D cities
Sep 17, 2026 · DreamFly: Causal Memory and Receding-Horizon Diffusion Planning for Aerial Vision-Language Navigation
- EP435: AI agents that learn from their misclicks
Sep 17, 2026 · Test-Time Self-Evolving GUI Visual Grounding via Reflection-Guided On-Policy Self-Distillation
- EP434: AI Panels Debate to Solve Medical Mysteries
Sep 16, 2026 · Social Chain of Thought: A Multi-Agent Architecture Grounded in Medical Differential Diagnosis Methodology
- EP433: Joint planning for mathematically guaranteed code
Sep 16, 2026 · P$^{3}$: Joint Program-and-Proof Planning for Verified Code Generation
- EP432: Motif 3 Replaces Brute Force with Specialization
Sep 15, 2026 · Motif 3: Technical Report
- EP431: Yale MoRSE ends AI agent redundancy
Sep 15, 2026 · MoRSE: Task-Oriented Multi-Agent System with Mixture of Role-Subtask Experts
- EP430: Khora Scales Real Time AI Hallucinated Worlds
Sep 14, 2026 · Population-Scalable Multi-Agent World Modeling
- EP429: AI agents redesigning their own software harnesses
Sep 14, 2026 · Hierarchical Self-Improvement: A Framework for Task-Specific Evolvable Agent Harnesses
- EP428: Tiny AI beats giants with silent logic
Sep 13, 2026 · Think Deep, Speak Once: Relit, A Recursive Latent Implicit Transformer Framework
- EP427: Replacing AI reasoning with distilled skills
Sep 13, 2026 · Reason Wide, Not Deep: Amortizing the Reasoning Premium into Distilled Skills
- EP426: HiLP enables long horizon AI planning
Sep 12, 2026 · Hierarchical Latent Prediction for Language Models
- EP425: AI agents playing actor and environment
Sep 12, 2026 · EnvACE: Internalizing Environment Dynamics via World Rehearsal for Agentic Reinforcement Learning
- EP424: Why Argus AI Thrives on Dead Ends
Sep 11, 2026 · Argus: A General-Purpose Agentic Runtime for Long-Horizon Reasoning
- EP423: Why Agentic AI breaks the datacenter
Sep 11, 2026 · Architectural Implications of Agentic AI Workflows
- EP422: Stopping spurious signals in AI distillation
Sep 10, 2026 · When Teachers Mislead: Spurious-Signal-Aware On-Policy Distillation
- EP421: Fixing AI Hallucinations with RAIL Principles
Sep 10, 2026 · The RAIL Principles for Neurosymbolic AI: Reasoning, Assurances, Interfacing and Learning
- EP420: Hijacking AI memory via factual injection
Sep 9, 2026 · MAFIA: Query-Only Memory Attacks via Probing and Factual Injection against Audited LLM Agents
- EP419: Ten Weeks of Autonomous AI Research
Sep 9, 2026 · Long-Horizon Autonomous Architecture Research with a Language-Model Agent: A Behavioural Case Study
- EP418: DeepVoyager-VL solves the visual search bottleneck
Sep 8, 2026 · DeepVoyager-VL: Incentivizing Vision-in-the-Loop Search for Long-Horizon Multimodal Agents
- EP417: AI agents replace human beta testers
Sep 8, 2026 · Can AI Agents Simulate A/B Test Outcomes? A Validation Framework for Agentic Experimentation
- EP416: How AdaThinkV stops AI overthinking video
Sep 7, 2026 · AdaThinkV: Adaptive Thinking for Token-Efficient Video Reasoning
- EP415: Fixing the AI granularity mismatch
Sep 7, 2026 · Where Reasoning Diverges: Localized Multi-Agent Debate
- EP414: Why context compaction breaks AI agents
Sep 6, 2026 · Context Compaction Theory
- EP413: Slashing AI latency with uncertainty repair
Sep 6, 2026 · CURE: Local Uncertainty Repair for Block-Parallel Speculative Decoding
- EP412: Thermodynamic Computing Solves the AI Bottleneck
Sep 5, 2026 · CN101 - A Digital Thermodynamic Computer for Generative AI
- EP411: How NeSyFS Gives AI Fast-Slow Thinking
Sep 5, 2026 · NeSyFS: A Neuro-symbolic Fast-Slow Thinking Framework for LLM Agent under Partial Observability
- EP410: How provenance laundering brainwashes AI
Sep 4, 2026 · Memory Provenance Laundering in LLM Agents: A Non-Amplification Firewall for Persistent Memory
- EP409: Robots That Dream Before They Move
Sep 4, 2026 · World Action Planner: Generalizable Decision-Making with Action-Conditioned World Models
- EP408: AI memory reconstructed not replayed
Sep 3, 2026 · MemHarness: Memory Is Reconstructed, Not Replayed
- EP407: How AI learns your teamwork capabilities
Sep 3, 2026 · Partner Capability Estimation for Task-Agnostic Adaptation in Ad-Hoc Teamwork
- EP406: Ending AI Groundhog Day With Living Harness
Sep 2, 2026 · Living-Harness Is an Interactive-Agent Evolver
- EP405: Are AI Agents Just Talking to Themselves
Sep 2, 2026 · Do Latent Channels Actually Communicate? A Causal Audit of Latent Multi-Agent LLM
- EP404: AI agents hide betrayal in Werewolf
Sep 1, 2026 · Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systems
- EP403: COVENANT keeps AI agents on the rails
Sep 1, 2026 · COVENANT: Natural-Language Workflow Compilation for Aligned Agent Execution
- EP402: Static baselines beat dynamic AI agents
Aug 31, 2026 · A Control System, a Dataset, and a Recipe for Making Frozen LLM Agents Learn a Domain
- EP401: Extracting Pure Reasoning From AI Giants
Aug 31, 2026 · From Proprietary to Open-
- EP400: Can GPT-5.1 understand the world
Aug 30, 2026 · Embodied GPT-5.1: Evidence of a World Model?
- EP399: Training AI to follow any reasoning workflow
Aug 30, 2026 · Training Language Models to Cooperate with Inference-Time Controllers
- EP398: Social deduction games teach AI creativity
Aug 29, 2026 · From RLVR to RLSVR: Task Transformation Induces Self-Verifiable Rewards for Open-Ended LLM Self-Improvement
- EP397: Teaching AI teams to focus
Aug 29, 2026 · Focus Is All You Need: Adaptive Goal-aware Attention Orchestration for Multi-Agent Graph Systems
- EP396: How numerical scores trigger AI reinforcement learning
Aug 28, 2026 · In-Context Learning as Implicit Policy Gradient
- EP395: ConsistencyGate stops AI memory contamination
Aug 28, 2026 · ConsistencyGate: Preventing Memory Contamination in LLM Agents via Self-Consistency Admission Control
- EP394: Agentic Context Management Beats Raw Compute
Aug 27, 2026 · Agentic Context Management: Solving Agent Memory and Cost by Treating Them as Lifecycle and Architecture Problems
- EP393: Why AREX agents audit their own research
Aug 27, 2026 · AREX: Towards a Recursively Self-Improving Agent for Deep Research
- EP392: Small models beat giants at malware analysis
Aug 26, 2026 · Small, Free, and Effective: Orchestrating Open-Weight Small Language Models to Outperform Single LLM for Malware Analysis
- EP391: Programmatic memory fixes AI context rot
Aug 26, 2026 · PRO-LONG: Programmatic Memory Enables Long-Horizon Reasoning
- EP390: EvoDRC solves microscopic silicon design errors
Aug 25, 2026 · EvoDRC: A Self-Evolving Agentic Framework for Automated DRC Violation Repair
- EP389: Solving the AI Memory Trilemma
Aug 25, 2026 · Supra Cognitive Modes: A Routed Architecture for Agent Memory
- EP388: Machine translation with latent reasoning loops
Aug 24, 2026 · LatentMT: Machine Translation with Latent Reasoning
- EP387: Shared libraries for disposable AI agents
Aug 24, 2026 · Knowledge-Centric Self-Improvement
- EP386: Infinite playable worlds on a single GPU
Aug 23, 2026 · ABot-World-0: Infinite Interactive World Rollout on a Single Desktop GPU
- EP385: AI self-correction can destroy correct answers
Aug 23, 2026 · Verify, Repair, Repeat, or Stop? Robust Stopping for Noisy Verify-Repair Loops in LLM Agents
- EP384: How AI can finally stop forgetting
Aug 22, 2026 · The Art of Not Forgetting
- EP383: Why AI Agents Disobey Their Own Logic
Aug 22, 2026 · Operational Hallucination and Safety Drift in AI Agents
- EP382: Etas The Native Language For AI Agents
Aug 21, 2026 · ETAS: An Effect-Typed Language for Agent Systems
- EP381: How AI finally learned to smell
Aug 21, 2026 · COLIP-2: Olfaction-Vision-Language Embeddings
- EP380: AI rewriting itself to solve formal math
Aug 20, 2026 · Self-Modifying Lean Proof Agents with Verifier-Grounded Benchmark Coevolution
- EP379: Smaller AI beats giants by thinking twice
Aug 20, 2026 · Reward-Driven LLM Agent Workflows: Synthesizing POMDP Routing and Self-Correction for Autonomous Decision-Making
- EP378: Giving AI a Silent Inner Monologue
Aug 19, 2026 · Training Continuous Chain of Thought Models: A Tale of Two Regimes
- EP377: PRIME Solves AI Curiosity Traps
Aug 19, 2026 · Principled Direction-Free Intrinsic Motivation through Model-Free Epistemic Free-Energy Estimators
- EP376: ToolVerse Teaches AI to Execute Complex Tasks
Aug 18, 2026 · ToolVerse: Unlocking Massive Environments and Long-Horizon Tasks for Agentic Reinforcement Learning
- EP375: M2GDT solves multimodal knowledge graph completion
Aug 18, 2026 · MGDT: MLLM-Guided Diffusion Transformer with Relation-Adaptive Mixture-of-Experts for Multimodal Knowledge Graph Completion
- EP374: TopoAgent Outperforms GPT-5 in Science
Aug 17, 2026 · TopoAgent: A Self-Evolving Topological Agent for Multimodal Scientific Reasoning
- EP373: Middle Layer Recurrence Fixes AI Amnesia
Aug 17, 2026 · T^2MLR: Transformer with Temporal Middle-Layer Recurrence
- EP372: How UrbanAgent profiles unseen cities
Aug 16, 2026 · Multi-Agent Collaborative Reasoning with Tool-Augmented Evidence for Urban Region Profiling
- EP371: Groc-PO Stops Multimodal AI Hallucinations
Aug 16, 2026 · Groc-PO: Grounded Context Preference Optimization for Truthful Multimodal LLMs
- EP370: SLEUTH fixes AI multi-hop reasoning failures
Aug 15, 2026 · Track, Rank, Crack: Epistemic Working Memory Scales Multi-Hop Reasoning in Language Agents
- EP369: How Atomic Units Scale Intelligence
Aug 15, 2026 · Atomic Units of X: The Compression Layer of Intelligence
- EP368: Samba Framework for Audio-Visual Navigation
Aug 14, 2026 · A Hybrid Mamba for Audio-Visual Navigation
- EP367: Why AI sounds so painfully corporate
Aug 14, 2026 · Optimization Is Not All You Need
- EP366: Autonomous AI Writes Its Own Hacking Tools
Aug 13, 2026 · Mako: A Self-Evolving Agentic Operating System (SE-AOS) for Autonomous Web Exploitation
- EP365: Smarter managers beat bigger AI brains
Aug 13, 2026 · Agentic Routing: The Harness-Native Data Flywheel
- EP364: Capability Trees for Scalable AI Agents
Aug 12, 2026 · A Formal Hierarchical Architecture for Agentic Orchestration with Stack-Based Execution and Lazy Discovery
- EP363: How Logos Architecture Stops AI Misevolution
Aug 12, 2026 · LOGOS: A Living Logic for AI Agent Teams That Evolve With Humans
- EP362: How Agentic-DPO fixes brittle AI agents
Aug 11, 2026 · Agentic-DPO: From Imitation to Agentic Policy Optimization on Expert Trajectories
- EP361: How Riemannian geometry fixes AI reasoning
Aug 11, 2026 · Beyond Euclidean Clipping: Overcoming Exploration Collapse in LLM RL via Riemannian Isometric Policy Optimization
- EP360: How ARMOR stops AI reasoning collapse
Aug 10, 2026 · ARMOR: Stabilizing On-Policy LLM RL with Off-Policy Anchor Samples
- EP359: Why your AI should forget
Aug 10, 2026 · Shared Selective Persistent Memory for Agentic LLM Systems
- EP358: Europe s Transparent Soofi S AI Blueprint
Aug 9, 2026 · A Sovereign, Open-Source Foundation Model for German and English
- EP357: Copying Smart Experts Makes AI Worse
Aug 9, 2026 · Compete Then Collaborate: Frontier AI Teachers Build a Verifiable Curriculum to Improve a Coding Student Beyond Imitation
- EP356: CMA solves the visual token explosion
Aug 8, 2026 · Cognitive-structured Multimodal Agent for Multimodal Understanding, Generation, and Editing
- EP355: RL builds compositional reasoning strategies
Aug 8, 2026 · RL Post-Training Builds Compositional Reasoning Strategies
- EP354: How AI Agents Code Their Own Habits
Aug 7, 2026 · From Atomic Actions to Standard Operating Procedures: Iterative Tool Optimization for Self-Evolving LLM Agents
- EP353: How IGRPO stops AI search distractions
Aug 7, 2026 · Information Gain-based Rollout Policy Optimization: An Adaptive Tree-Structured Rollout Approach for Multi-Turn LLM Agents
- EP352: Hidden states predict AI agent failure
Aug 6, 2026 · Doomed from the Start: Early Abort of LLM Agent Episodes via a Recall-Controlled Probe Cascade
- EP351: Direct-OPD slashes AI reasoning compute costs
Aug 6, 2026 · Weak-to-Strong Generalization via Direct On-Policy Distillation
- EP350: Training AI agents without live environments
Aug 5, 2026 · Multi-Turn On-Policy Distillation with Prefix Replay
- EP349: Fixing AI judges with continuous verification
Aug 5, 2026 · LLM-as-a-Verifier: A General-Purpose Verification Framework
- EP348: Building AI agents like living cells
Aug 4, 2026 · Biological Motifs for Agentic Control
- EP347: Compiling AI into Permanent Free Skills
Aug 4, 2026 · Auto: The AGI Compiler
- EP346: Teaching small AI to ignore teachers
Aug 3, 2026 · Reward-Gated On-Policy Distillation
- EP345: AI agents retry from pivotal mistakes
Aug 3, 2026 · Agent Reinforcement Learning via Pivotal-Aware Self-Feedback Retry
- EP344: AI predicts tool calls to skip waiting
Aug 2, 2026 · SPORK: Self-Speculative Forking to Accelerate Agentic LLM Inference
- EP343: How AI agents escape infinite loops
Aug 2, 2026 · No Time Like the Present: Agentic Test-Time Training for LLM Agents
- EP342: Why process rubrics triple AI accuracy
Aug 1, 2026 · SkillCoach: Self-Evolving Rubrics for Evaluating and Enhancing Agentic Skill-Use
- EP341: Gemma 4 brings thinking mode to laptops
Aug 1, 2026 · Gemma 4 Technical Report
- EP340: AI Models Prove Opposite Scientific Truths
Jul 31, 2026 · The Agentic Garden of Forking Paths
- EP339: How AI Safely Rewrites Its Own Code
Jul 31, 2026 · Self-Evolving Agents with Anytime-Valid Certificates
- EP338: DiscoPER conducts autonomous science via reflection
Jul 30, 2026 · Autonomous Scientific Discovery via Iterative Meta-Reflection
- EP337: Why AI Agents Fail in Silence
Jul 30, 2026 · Ask the World Before Acting: Budgeted Environment Probing for World-Model Calibration
- EP336: ACE fixes the AI goldfish memory problem
Jul 29, 2026 · ACE: Pluggable Adaptive Context Elasticizer across Agents
- EP335: How AI agents learn from failure
Jul 29, 2026 · Self-Evolving World Models for LLM Agent Planning
- EP334: Fixing AI Hallucinations With Process Rewards
Jul 28, 2026 · SEVA: Self-Evolving Verification Agent with Process Reward for Fact Attribution
- EP333: Logic not length makes AI smarter
Jul 28, 2026 · Does Verbose Chain-of-Thought Really Help? In-Distribution Evidence that Content, Not Length, Matters
- EP332: AI Architects Designing Better Embodied Agents
Jul 27, 2026 · Automating the Design of Embodied AgentArchitectures
- EP331: Internalizing AI debate with Mixture of Debaters
Jul 27, 2026 · Mixture of Debaters: Learn to Debate at Architectural Level in Multi-Agent Reasoning
- EP330: AI agents audit 10,000 page nuclear reports
Jul 26, 2026 · LLM-Guided Planning for Multi-hop Reasoning over Multimodal Nuclear Regulatory Documents
- EP329: Teaching AI to forget the right things
Jul 26, 2026 · Coverage-Driven KV Cache Eviction for Efficient and Improved Inference of LLM
- EP328: FlowWM and branching futures
Jul 25, 2026 · Flow Matching in Feature Space for Stochastic World Modeling
- EP327: Why Chatbot Safety Training Backfires for Agents
Jul 25, 2026 · Agent Safety Is Action Alignment
- EP325: Why robots have too much brain
Jul 24, 2026 · Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models?
- EP324: JERP synchronizes AI rules and neural weights
Jul 23, 2026 · Joint Learning of Experiential Rules and Policies for Large Language Model Agents
- EP323: Giving AI Einstein s visual imagination
Jul 23, 2026 · Einstein World Models
- EP322: Why cliff tokens break AI math
Jul 22, 2026 · Cliff Tokens: Identifying Single-Token Failure Triggers in LLM Mathematical Reasoning
- EP321: Measuring AI intelligence in bits
Jul 22, 2026 · Agentic System as Compressor: Quantifying System Intelligence in Bits
- EP320: Universal AI is mathematically impossible
Jul 21, 2026 · World Models in Pieces: Structural Certification for General Agents
- EP319: How TRUSTMEM Fixes Broken AI Memory
Jul 21, 2026 · TRUSTMEM: Learning Trustworthy Memory Consolidation for LLM Agents with Long-Term Memory
- EP318: Open Data Recipes for AI Agents
Jul 20, 2026 · OpenThoughts-Agent: Data Recipes for Agentic Models
- EP317: The Architecture Of Genuine Artificial Agency
Jul 20, 2026 · Critique of Agent Model
- EP316: Teaching robotaxis the biological urge to survive
Jul 19, 2026 · Active Inference as the Test-Time Scaling Law for Physical AI Agents
- EP315: Teaching robots to think like scientists
Jul 19, 2026 · Self-Evolving Cognitive Framework via Causal World Modeling for Embodied Scientific Intelligence
- EP314: Why AI hacks its own geometry
Jul 18, 2026 · All Routes Lead to Collapse
- EP313: How ARTS reasons through its own failures
Jul 18, 2026 · Learning the ARTS of Search for Automated Discovery
- EP312: BioMatrix translates English to 3D biology
Jul 17, 2026 · BioMatrix: Towards a Comprehensive Biological Foundation Model Spanning the Modality Matrix of Sequences, Structures, and Language
- EP311: Why AI Teams Hallucinate Together
Jul 17, 2026 · Hallucination as Context Drift: Synchronization Protocols for Multi-Agent LLM Systems
- EP310: Why AI Breaks While Fixing Itself
Jul 16, 2026 · Denoising Iterative Self-Correction: Structured Verification Loops for Reliable LLM Reasoning
- EP309: AutoRAS builds self-healing AI agent networks
Jul 16, 2026 · AutoRAS: Learning Robust Agentic Systems with Primitive Representations
- EP308: Giving AI Agents Mathematical Muscle Memory
Jul 15, 2026 · SoftSkill: Behavioral Compression for Contextual Adaptation
- EP307: AI agents now train physical robots autonomously
Jul 15, 2026 · ENPIRE: Agentic Robot Policy Self-Improvement in the Real World
- EP306: AIs that engineer their own pipelines
Jul 14, 2026 · FAPO: Fully Autonomous Prompt Optimization of Multi-Step LLM Pipelines
- EP305: Mathematical guardrails for autonomous AI agents
Jul 14, 2026 · Deontic Policies for Runtime Governance of Agentic AI Systems
- EP304: MagicSim Bridges AI and Physics
Jul 13, 2026 · MagicSim: A Unified Infrastructure for Executable Embodied Interaction
- EP303: How MODE-RAG stops AI video lies
Jul 13, 2026 · MODE-RAG: Manifold Outlier Diagnosis and Energy-based Retrieval-Augmented Generation Evaluation
- EP302: Transferable interaction patterns for web agents
Jul 12, 2026 · Beyond Domains: Reusing Web Skills via Transferable Interaction Patterns
- EP301: VeriGraph Makes AI Data Analysis Verifiable
Jul 12, 2026 · VeriGraph: Towards Verifiable Data-Analytic Agents
- EP300: Tensors prevent multi-agent LLM collisions
Jul 11, 2026 · Tensor-Coord: Algebraic Decomposition of Joint Plan Tensors for Conflict-Free Multi-Agent LLM Planning
- EP299: STRIDE grades the AI scratchpad
Jul 11, 2026 · STRIDE: Strategic Trajectory Reasoning via Discriminative Estimation for Verifiable Reinforcement Learning
- EP298: LLM-as-Code Fixes Unreliable AI Agents
Jul 10, 2026 · LLM-as-Code Agentic Programming for Agent Harness
- EP297: How T-Mem fixes the associative blind spot
Jul 10, 2026 · T-Mem: Memory That Anticipates, Not Archives
- EP296: Stop parallel AI agents from crashing production
Jul 9, 2026 · CoAgent: Concurrency Control for Multi-Agent Systems
- EP295: Ending agent sprawl with canonical code
Jul 9, 2026 · No Accidental Software Agent First Canonical Code for Human Code Entropy Reduction and 30 to 500 times Lower Frontier Model Requirements
- EP294: Why AI agents second-guess their success
Jul 8, 2026 · Closing the Reflection Gap: A Free Calibration Bonus for Agentic RL
- EP293: Grading AI blueprints with Orch-RM
Jul 8, 2026 · Reward Modeling for Multi-Agent Orchestration
- EP292: Agents-K1 turns AI into research scientists
Jul 7, 2026 · Agents-K1: Towards Agent-native Knowledge Orchestration
- EP291: Ouroboros-Spatial Outperforms AI Giants in 3D
Jul 7, 2026 · Ouroboros-Spatial: Closing the Data-Model Loop for Spatial Reasoning
- EP290: How Knowledge Graphs Fix Multi-Hop Reasoning
Jul 6, 2026 · LLMs+Graphs: Toward Graph-Native, Synergistic AI Systems
- EP289: Runtime governance for autonomous AI agents
Jul 6, 2026 · A Five-Plane Reference Architecture for Runtime Governance of Production AI Agents
- EP288: Test-Time Training shatters quadratic sampling limits
Jul 5, 2026 · The Power of Test-Time Training for Approximate Sampling
- EP287: Small models beat GPT-4o with Role-Agent
Jul 5, 2026 · Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution
- EP286: ReasonAlloc Solves the AI Memory Bottleneck
Jul 4, 2026 · ReasonAlloc: Hierarchical Decoding-Time KV Cache Budget Allocation for Reasoning Models
- EP285: How SkeMex builds medical AI intuition
Jul 4, 2026 · Experience Makes Skillful: Enabling Generalizable Medical Agent Reasoning via Self-Evolving Skill Memory
- EP284: Compressing massive context into soft tokens
Jul 3, 2026 · End-to-End Context Compression at Scale
- EP283: Aligning AI planners with tool capabilities
Jul 3, 2026 · Capability-Aligned Hierarchical Learning for Tool-Augmented LLMs
- EP282: AI gladiators training in shopping arenas
Jul 2, 2026 · Bittensor Agent Arenas as a Trajectory Primitive: Distilling a Shopping Agent from ShoppingBench Subnet Traces
- EP281: Restoring plasticity to over-trained AI
Jul 2, 2026 · When RL Fails after SFT: Rejuvenating Model Plasticity for Robust SFT-to-RL Handoff
- EP280: Trajectory Refined Distillation Fixes AI Reasoning
Jul 1, 2026 · Trajectory-Refined Distillation
- EP279: Ending AI amnesia with strategy cards
Jul 1, 2026 · ConMem: Structured Memory-Guided Adaptation in Training-Free Multi-Agent Systems
- EP278: Hacking AI Agents with Fake Errors
Jun 30, 2026 · VATS: Exploiting Implicit Authority in Error-Path Injection via Systematic Mutation
- EP277: AI quorums stop cloud infrastructure failures
Jun 30, 2026 · Semantic Quorum Assurance: Collective Certification for Non-Deterministic AI Infrastructure
- EP276: ThinkBooster scales LLM reasoning at test time
Jun 29, 2026 · ThinkBooster: A Unified Framework for Seamless Test-Time Scaling of LLM Reasoning
- EP275: AI Agents Building Their Own Coding Curriculum
Jun 29, 2026 · Socratic-SWE: Self-Evolving Coding Agents via Trace-Derived Agent Skills
- EP274: Knowledge graphs fix AI memory loss
Jun 28, 2026 · TokenMizer: Graph-Structured Session Memory for Long-Horizon LLM Context Management
- EP273: Why agents make code disposable
Jun 28, 2026 · The End of Software Engineering: How AI Agents Are Fundamentally Restructuring the Software Paradigm
- EP272: AI rewiring its own brain live
Jun 27, 2026 · Scaling Self-Evolving Agents via Parametric Memory
- EP271: Steer locked AI with Agentic Monte Carlo
Jun 26, 2026 · Agentic Monte Carlo: Simulating Reinforcement Learning for Black-Box Agents
- EP270: AI agents building their own reasoning tools
Jun 26, 2026 · Inducing Reasoning Primitives from Agent Traces
- EP269: Securing AI Agents with Agent libOS
Jun 25, 2026 · Agent libOS: A Library-OS-Inspired Runtime for Long-Running, Capability-Controlled LLM Agents
- EP268: How OpenWebRL masters the live web
Jun 25, 2026 · OpenWebRL: Demystifying Online Multi-turn Reinforcement Learning for Visual Web Agents
- EP267: AI Agents That Update Their Own Imagination
Jun 24, 2026 · COMAP: Co-Evolving World Models and Agent Policies for LLM Agents
- EP266: AI agents learn to think without words
Jun 24, 2026 · Adaptive Latent Agentic Reasoning
- EP265: How AI agents rewrite their own tools
Jun 23, 2026 · SkillSmith: Co-Evolving Skills and Tools for Self-Improving Agent Systems
- EP264: Science Earth and Planet Scale AI Discovery
Jun 23, 2026 · Science Earth: Towards A Planet-Scale Operating System for AI-Native Scientific Discovery
- EP263: How POPO ends AI training waste
Jun 22, 2026 · RLVR without Ineffective Samples: Group Prioritized Off-Policy Optimization for LLM Reasoning
- EP262: Web agents that learn from failure
Jun 22, 2026 · Learning to Adapt: Self-Improving Web Agent via Cognitive-Aware Exploration
- EP261: EchoRL turns hesitation into genius
Jun 21, 2026 · EchoRL: Reinforcement Learning via Rollout Echoing
- EP260: GrepSeek brings Unix precision to AI
Jun 21, 2026 · GrepSeek: Training Search Agents for Direct Corpus Interaction
- EP259: The ESPO Kill Switch For AI Reasoning
Jun 20, 2026 · ESPO: Early-Stopping Proximal Policy Optimization
- EP258: TRACER teaches AI to stay silent
Jun 20, 2026 · TRACER: Turn-level Regret Matching with Inner Reinforcement Credit for Cooperative Multi-LLM Reasoning
- EP257: How planning wakes up deep AI layers
Jun 19, 2026 · Do Agents Think Deeper? A Mechanistic Investigation of Layer-Wise Dynamics in Sequential Planning
- EP256: Teaching AI to Doubt Its Own Answers
Jun 19, 2026 · Confidence-Orchestrated Self-Evolution against Uncertain LLM Feedback
- EP255: MUSE-Autoskill creates self-evolving AI agents
Jun 18, 2026 · MUSE-Autoskill: Self-Evolving Agents via Skill Creation, Memory, Management, and Evaluation
- EP254: Why Innovation Guarantees AI Hallucination
Jun 18, 2026 · Innovation: An Almost Characterization of Hallucination
- EP253: MACA optimizes AI agent coordination
Jun 17, 2026 · Multi-Agent Coordination Adaptation via Structure-Guided Orchestration
- EP252: How batch sizes sharpen AI reasoning
Jun 17, 2026 · How Should LLMs Consume High-Quality Data? Optimal Data Scheduling via Quality-Aware Functional Scaling Laws
- EP251: How SR2AM stops AI overthinking
Jun 16, 2026 · Efficient Agentic Reasoning Through Self-Regulated Simulative Planning
- EP250: Compiling agent workflows into model weights
Jun 16, 2026 · Compiling Agentic Workflows into LLM Weights: Near-Frontier Quality at Two Orders of Magnitude Less Cost
- EP249: Mem-pi fixes AI amnesia with generative memory
Jun 15, 2026 · Mem-π: Adaptive Memory through Learning When and What to Generate
- EP248: 10x Faster AI Agents with JIT Compilation
Jun 15, 2026 · Agent JIT Compilation for Latency-Optimizing Web Agent Planning and Scheduling
- EP247: PEEK Cures AI Goldfish Memory
Jun 14, 2026 · PEEK: Context Map as an Orientation Cache for Long-Context LLM Agents
- EP246: Replacing AI manuals with programmable runtimes
Jun 14, 2026 · Formal Skill: Programmable Runtime Skills for Efficient and Accurate LLM Agents
- EP245: The Geometric Shape of AI Reasoning
Jun 13, 2026 · A Measure-Theoretic Analysis of Reasoning: Structural Generalization and Approximation Limits
- EP244: Training decentralized AI through private handoffs
Jun 13, 2026 · Learning to Hand Off: Provably Convergent Workflow Learning under Interface Constraints
- EP243: Breaking the AI data wall with SYNPRO
Jun 12, 2026 · Generating Pretraining Tokens from Organic Data for Data-Bound Scaling
- EP242: Ending AI Amnesia with Experience Graphs
Jun 12, 2026 · EXG: Self-Evolving Agents with Experience Graphs
- EP241: Accelerating game theory with linear algebra
Jun 11, 2026 · Parallelizing Counterfactual Regret Minimization
- EP240: Small AI agents beat giants with Orchard
Jun 11, 2026 · Orchard: An Open-Source Agentic Modeling Framework
- EP239: The shift from chatbots to AI societies
Jun 10, 2026 · Beyond Individual Intelligence: Surveying Collaboration, Failure Attribution, and Self-Evolution in LLM-based Multi-Agent Systems
- EP238: SepsisAgent outperforms clinicians using clinical world models
Jun 10, 2026 · Agentifying Patient Dynamics within LLMs through Interacting with Clinical World Model
- EP237: Why AI agents must map before acting
Jun 9, 2026 · MAP: A Map-then-Act Paradigm for Long-Horizon Interactive Agent Reasoning
- EP236: AI agents rewriting their own code
Jun 9, 2026 · Harnessing Agentic Evolution
- EP235: How SAGE Fixes AI Memory
Jun 8, 2026 · SAGE: A Self-Evolving Agentic Graph-Memory Engine for Structure-Aware Associative Memory
- EP234: FATE fixes safe but useless AI agents
Jun 8, 2026 · On-Policy Self-Evolution via Failure Trajectories for Agentic Safety Alignment
- EP233: Fixing AI memory with backward chaining
Jun 7, 2026 · Goal-Oriented Reasoning for RAG-based Memory in Conversational Agentic LLM Systems
- EP232: Why AI agents lie to fit in
Jun 7, 2026 · The Bystander Effect in Multi-Agent Reasoning: Quantifying Cognitive Loafing in Collaborative Interactions
- EP231: Amazon PIVOT solves the AI execution gap
Jun 6, 2026 · PIVOT: Bridging Planning and Execution in LLM Agents via Trajectory Refinement
- EP230: DeepRefine fixes messy AI knowledge bases
Jun 6, 2026 · DeepRefine: Agent-Compiled Knowledge Refinement via Reinforcement Learning
- EP229: Ending the AI verbosity tax with LEAD
Jun 5, 2026 · LEAD: Length-Efficient Adaptive and Dynamic Reasoning for Large Language Models
- EP228: Why self-evolving AI forgets basic tasks
Jun 5, 2026 · Do Self-Evolving Agents Forget? Capability Degradation and Preservation in Lifelong LLM Agent Adaptation
- EP227: FlowAgent fixes the AI tool bottleneck
Jun 4, 2026 · Tools as Continuous Flow for Evolving Agentic Reasoning
- EP226: MELT Decouples AI Reasoning from Memory
Jun 4, 2026 · Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models
- EP225: Turning AI into its own lie detector
Jun 3, 2026 · Logic-Regularized Verifier Elicits Reasoning from LLMs
- EP224: Soft-Hamiltonian world models for robust planning
Jun 3, 2026 · HaM-World: Soft-Hamiltonian World Models with Selective Memory for Planning
- EP223: UNO-ORCHESTRA Slashes AI Costs via Selective Delegation
Jun 2, 2026 · Uno-Orchestra: Parsimonious Agent Routing via Selective Delegation
- EP222: Gyan Beats GPT-4o Without Using GPUs
Jun 2, 2026 · Gyan: An Explainable Neuro-Symbolic Language Model
- EP221: ScrapMem Mimics Human Memory Through Forgetting
Jun 1, 2026 · ScrapMem: A Bio-inspired Framework for On-device Personalized Agent Memory via Optical Forgetting
- EP220: How PARSE Makes AI Four Times Faster
Jun 1, 2026 · Parallel Prefix Verification for Speculative Generation
- EP219: OpenSeeker V2 Shatters The AI Compute Myth
May 31, 2026 · OpenSeeker-v2: Pushing the Limits of Search Agents with Informative and High-Difficulty Trajectories
- EP218: JoyAI-Image Solves AI 3D Geometry Errors
May 31, 2026 · Awaking Spatial Intelligence in Unified Multimodal Understanding and Generation
- EP217: Why forced compliance triggers metacognitive collapse
May 30, 2026 · The Compliance Trap: How Structural Constraints Degrade Frontier AI Metacognition Under Adversarial Pressure
- EP216: Shadow memory stops long horizon AI heists
May 30, 2026 · MAGE: Safeguarding LLM Agents against Long-Horizon Threats via Shadow Memory
- EP215: Finding specialized AI agents in milliseconds
May 29, 2026 · GRAIL: A Deep-Granularity Hybrid Resonance Framework for Real-Time Agent Discovery via SLM-Enhanced Indexing
- EP214: ARISE Maps Data Flow For AI Agents
May 29, 2026 · ARISE: A Repository-level Graph Representation and Toolset for Agentic Fault Localization and Program Repair
- EP213: Why AI agents fail at negotiation
May 28, 2026 · Talk is Cheap, Communication is Hard: Dynamic Grounding Failures and Repair in Multi-Agent Negotiation
- EP212: Sheaf Geometry Fixes Robot Logic
May 28, 2026 · Sheaf-Theoretic Planning: A Categorical Foundation for Resilient Multi-Agent Autonomous Systems
- EP211: SciResearcher turns AI into a scientific detective
May 27, 2026 · SciResearcher: Scaling Deep Research Agents for Frontier Scientific Reasoning
- EP210: AI that rewrites its own logic
May 27, 2026 · Lifting Traces to Logic: Programmatic Skill Induction with Neuro-Symbolic Learning for Long-Horizon Agentic Tasks
- EP209: Fixing AI agent memory with SAGA
May 26, 2026 · SAGA: Workflow-Atomic Scheduling for AI Agent Inference on GPU Clusters
- EP208: Bayesian Orchestration for Overconfident AI Agents
May 26, 2026 · Position: agentic AI orchestration should be Bayes-consistent
- EP207: Robots learn the math of anticipation
May 25, 2026 · PRTS: A Primitive Reasoning and Tasking System via Contrastive Representations
- EP206: ObjectGraph replaces Markdown for AI agents
May 25, 2026 · ObjectGraph: From Document Injection to Knowledge Traversal -- A Native File Format for the Agentic Era
- EP205: Qiushi AI Discovers Optical Computing Hardware
May 24, 2026 · End-to-end autonomous scientific discovery on a real optical platform
- EP204: Solving the AI compositionality crisis
May 24, 2026 · AGEL-Comp: A Neuro-Symbolic Framework for Compositional Generalization in Interactive Agents
- EP203: How AI Agents Trade Real Money
May 23, 2026 · Operating-Layer Controls for Onchain Language-Model Agents Under Real Capital
- EP202: Why ADEMA AI Never Loses The Plot
May 23, 2026 · ADEMA: A Knowledge-State Orchestration Architecture for Long-Horizon Knowledge Synthesis with LLMAgents
- EP201: Nautile-370M solves AI memory bottlenecks
May 22, 2026 · Nautile-370M: Spectral Memory Meets Attention in a Small Reasoning Model
- EP200: Kwai Summary Attention and the memory wall
May 22, 2026 · Kwai Summary Attention Technical Report
- EP199: Separation of Powers for AI Safety
May 21, 2026 · Structural Enforcement of Goal Integrity in AI Agents via Separation-of-Powers Architecture
- EP198: AI masters StarCraft using chat logs
May 21, 2026 · DLM: Unified Decision Language Models for Offline Multi-Agent Sequential Decision Making
- EP197: Teaching AI Agents to Plan Like Humans
May 20, 2026 · From Coarse to Fine: Self-Adaptive Hierarchical Planning for LLM Agents
- EP196: Forcing AI to Prove Its Logic
May 20, 2026 · ArgRE: Formal Argumentation for Conflict Resolution in Multi-Agent Requirements Negotiation
- EP195: How tool attention ends the tools tax
May 19, 2026 · Tool Attention Is All You Need: Dynamic Tool Gating and Lazy Schema Loading for Eliminating the MCP/Tools Tax in Scalable Agentic Workflows
- EP194: AI coding through mental simulation
May 19, 2026 · DryRUN: On the Role of Public Tests in LLM-Driven Code Generation
- EP193: AI image generators master physical reality
May 18, 2026 · Image Generators are Generalist Vision Learners
- EP192: Fixing AI memory with knowledge graphs
May 18, 2026 · Automatic Ontology Construction Using LLMs as an External Layer of Memory, Verification, and Planning for Hybrid Intelligent Systems
- EP191: Why AI Agents Blame Each Other
May 17, 2026 · Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment
- EP190: [OLLM] Replacing AI dice rolls with ten lanes
May 17, 2026 · OLLM: Options-based Large Language Models
- EP189: How Sessa architecture fixes AI amnesia
May 16, 2026 · Sessa: Selective State Space Attention
- EP188: [Agent-World] AI Building Its Own Training Worlds
May 16, 2026 · Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
- EP187: Hive fixes multi-agent AI memory bottlenecks
May 15, 2026 · Hive: A Multi-Agent Infrastructure for Algorithm- and Task-Level Scaling
- EP186: Harness engineering for near perfect small models
May 15, 2026 · Compiling Deterministic Structure into SLM Harnesses
- EP185: Why AI architecture fails at logic
May 14, 2026 · The Topological Trouble With Transformers
- EP184: Defeating the AI consensus trap
May 14, 2026 · The Consensus Trap: Rescuing Multi-Agent LLMs from Adversarial Majorities via Token-Level Collaboration
- EP183: AI coding agents cheat with keywords
May 13, 2026 · Neurosymbolic Repo-level Code Localization
- EP182: AI logic is its weakest link
May 13, 2026 · Structured Abductive-Deductive-Inductive Reasoning for LLMs via Algebraic Invariants
- EP181: Small models beating GPT-5 with logic
May 12, 2026 · SGA-MCTS: Decoupling Planning from Execution via Training-Free Atomic Experience Retrieval
- EP180: How AI agents rewrite their code
May 11, 2026 · Autogenesis: A Self-Evolving Agent Protocol
- EP179: AIBuildAI Builds New AI Models From Scratch
May 11, 2026 · AIBuildAI: An AI Agent for Automatically Building AI Models
- EP178: AI agents reaching silent latent consensus
May 10, 2026 · Bridging MARL to SARL: An Order-Independent Multi-Agent Transformer via Latent Consensus
- EP177: CAPO math stops overconfident AI lies
May 9, 2026
- EP176: Trigonometry fixes the AI memory bottleneck
May 8, 2026
- EP175: How AI models teach themselves reasoning
May 7, 2026
- EP174: 1-bit Bonsai brings powerful AI offline
May 6, 2026
- EP173: AI models diagnosing diseases from blank scans
May 5, 2026
- EP172: How HyperAgents rewrite their own code
May 4, 2026
- EP171: Helium makes AI agent workflows 40x faster
May 3, 2026
- EP170: Qwen3.5 Multimodal Agent
May 2, 2026
- EP169: Cybersecurity Risks of Autonomous AI Agents
May 1, 2026
- EP168: Turning AI Agents into Mathematical Functions
Apr 30, 2026
- EP167: Why AI models ignore visual evidence
Apr 29, 2026
- EP166: The Auton solution to the integration paradox
Apr 28, 2026
- EP165: Translating hidden AI logic into English
Apr 27, 2026
- EP164: [LACONIC] Teaching AI to stop overthinking
Apr 26, 2026
- EP163: Why AI Models Only Remember Five Percent
Apr 25, 2026
- EP162: AI agents beat humans with malicious skills
Apr 24, 2026
- EP161: Small AI Judges Beat Massive Coding Giants
Apr 23, 2026
- EP160: [AgentSys] Securing AI agents with hierarchical memory
Apr 22, 2026
- EP159: Brute force scale dominates the AI frontier
Apr 21, 2026
- EP158: The hidden blind spots of AI logic
Apr 20, 2026
- EP157: [AgentHeLLM] Protecting drivers from hijacked vehicle AI
Apr 19, 2026
- EP156: [Uncertainty Quantification] How AI Agents Know They Are Guessing
Apr 18, 2026
- EP155: [Agentic Proposing] Small models beat giants with logic bricks
Apr 17, 2026
- EP154: [FS-Researcher] Giving AI agents a file system
Apr 16, 2026
- EP153: [SERA] Training AI coding agents on untested code
Apr 15, 2026
- EP152: DeepVerifier forces AI to check its work
Apr 14, 2026
- EP151: [MagicGUI-RMS] AI agents that think before they click
Apr 13, 2026
- EP150: The Leap to Autonomous Agentic Reasoning
Apr 12, 2026
- EP149: [IDRBench] Interactive AI beats lone wolf models
Apr 11, 2026
- EP148: How AI masters math through self-correction
Apr 10, 2026
- EP147: [DeepSynth-Eval] AI fails at deep research synthesis
Apr 9, 2026
- EP146: How InfiAgent solves the AI memory bottleneck
Apr 8, 2026
- EP145: [LongDA] Why smart AI fails at messy data
Apr 7, 2026
- EP144: [Evo-Memory] Building AI agents with self-evolving memory.
Apr 6, 2026
- EP143: Your AI will blackmail you to survive
Apr 5, 2026
- EP142: [DR-Arena] A ruthless arena for deep research agents
Apr 4, 2026
- EP141: [AIRS-Bench] AI agents beat human research benchmarks
Apr 3, 2026
- EP140: [LeWorldModel] AI learns physics on one GPU
Apr 2, 2026
- EP139: Mamba-3 Fixes the Transformer Memory Bottleneck
Apr 1, 2026
- EP138: [Mamba-2] Transformers and SSMs Are the Same Engine
Mar 31, 2026
- EP137: Attention Residuals Solve the LLM Depth Bottleneck
Mar 30, 2026
- EP136: Modular skills for autonomous AI agents
Mar 29, 2026
- EP135: [SoK] Curing AI Amnesia with Agentic Skills
Mar 28, 2026
- EP134: Autonomous AI squads building software
Mar 27, 2026
- EP133: RelayLLM Slashes AI Costs With Collaborative Decoding
Mar 26, 2026
- EP132: How Autonomous LLM Agents Actually Work
Mar 25, 2026
- EP131: MUSE creates self evolving AI agents
Mar 24, 2026
- EP130: [GAP] Graph-based planning for faster AI agents
Mar 23, 2026
- EP129: Why AI agents fail half the time
Mar 22, 2026
- EP128: MCP-Zero lets AI find its own tools
Mar 21, 2026
- EP127: Why tool use makes AI less intelligent
Mar 20, 2026
- EP126: OrcaLoca locates bugs in massive codebases
Mar 19, 2026
- EP125: Why AI Needs an Agent Computer Interface
Mar 18, 2026
- EP124: FRIDAY the AI that runs your computer
Mar 17, 2026
- EP123: MemGPT Turns LLMs into Operating Systems
Mar 16, 2026
- EP122: The Four Pillars of LLM Autonomous Agents
Mar 15, 2026
- EP121: How ToolLLaMA mastered 16000 real world APIs
Mar 14, 2026
- EP120: How Reflexion agents learn through verbal feedback
Mar 13, 2026
- EP119: HuggingGPT Turns LLMs Into AI Managers
Mar 12, 2026
- EP118: The AI Memory Wall Crisis
Mar 11, 2026
- EP117: AI agents learn through textual reflection
Mar 11, 2026
- EP116: Why AI struggles with empathy and interruptions
Mar 10, 2026
- EP115: Dr.LLM brings dynamic depth to AI
Mar 9, 2026
- EP114: FlashAttention-4 Solves Blackwell Hardware Bottlenecks
Mar 7, 2026
- EP113: How FlashAttention-3 Doubles H100 Speed
Mar 7, 2026
- EP112: GPT 5.4 Outperforms Human Professionals
Mar 7, 2026
- EP111: Claude Opus 4.6 Runs Businesses and Catches Manipulation
Mar 7, 2026
- EP110: Single agents beat expensive multi agent teams
Mar 5, 2026
- EP109: The Rise of Agentic Reasoning
Mar 4, 2026
- EP108: GPT-5 Can Lie and Play Dumb
Mar 1, 2026
- EP107: DeepMind’s SIMA 2 Masters Unseen Video Games
Mar 1, 2026
- EP106: Fixing AI Agents With Symbolic Guardrails
Mar 1, 2026
- EP105: iStar Autonomous Agents Grading Their Own Homework
Mar 1, 2026
- EP104: WebExplorer Beats Giants at Web Research
Mar 1, 2026
- EP103: Why AI Agents Think Themselves To Death
Mar 1, 2026
- EP102: Gemini 2.5 Thinks Before It Speaks
Mar 1, 2026
- EP101: Kimi k1.5 Breaks the AI Data Wall
Mar 1, 2026
- EP100: Meta's Llama 4 Herd Ends Monolithic Models
Mar 1, 2026
- EP099: Is AI Thinking Just Expensive Noise
Mar 1, 2026
- EP098: OpenAI o3 Hacked Its Own Grading System
Mar 1, 2026
- EP097: DeepSeek R1 Taught Itself to Reason
Mar 1, 2026
- EP096: Gemini 1.5 Pro's 10 Million Token Window
Mar 1, 2026
- EP095: Microsoft Phi-4 Beats Giants With Synthetic Data
Mar 1, 2026
- EP094: DeepSeek-V3 Rivals GPT-4 for $6 Million
Mar 1, 2026
- EP093: How OpenAI o1 Cracked the Strawberry Cipher
Mar 1, 2026
- EP092: BitNet b1.58 Replaces Multiplication With Addition
Mar 1, 2026
- EP091: Qwen 2.5 Beats Llama With Synthetic Data
Mar 1, 2026
- EP090: Pixtral 12B Beats Llama With Better Eyesight
Mar 1, 2026
- EP089: Qwen2-VL Gives AI Native Eyesight
Mar 1, 2026
- EP088: Qwen2 Beats Llama-3 Through Data Quality
Mar 1, 2026
- EP087: Meta's Chameleon Unifies Text and Images
Mar 1, 2026
- EP086: DeepSeek-V2 Breaks The Impossible Triangle
Mar 1, 2026
- EP085: Aya 23 Breaks The Curse Of Multilinguality
Mar 1, 2026
- EP084: Microsoft Phi-3 Fits Supercomputing in Your Pocket
Mar 1, 2026
- EP083: How Meta Engineered the Llama 3 Herd
Mar 1, 2026
- EP082: Command R Plus The Verifiable Enterprise Agent
Mar 1, 2026
- EP081: Replacing MLPs With Interpretable KANs
Mar 1, 2026
- EP080: Jamba Hybrid Solves Transformer Memory Limits
Mar 1, 2026
- EP079: DBRX Beats GPT-3.5
Mar 1, 2026
- EP078: Claude 3 Knew It Was Being Tested
Feb 28, 2026
- EP077: Google Squeezes Gemini Into Your Laptop
Feb 28, 2026
- EP076: OLMo Cracks Open the AI Black Box
Feb 28, 2026
- EP075: Microsoft Phi Beats Giants With Synthetic Textbooks
Feb 28, 2026
- EP074: How Gemini Beat Human Experts
Feb 28, 2026
- EP073: Mixtral 8x7B Sparse Experts Beat Giants
Feb 28, 2026
- EP072: Mamba Solves The Transformer's Fatal Flaw
Feb 28, 2026
- EP071: How Zephyr-7B Beat Llama-70B
Feb 28, 2026
- EP070: Mistral 7B Beats Llama 2 13B
Feb 28, 2026
- EP069: Alibaba's Qwen Specialized Models Beat Generalists
Feb 28, 2026
- EP068: vLLM Fixes the KV Cache Bottleneck
Feb 28, 2026
- EP067: FlashAttention-2 Unlocks Massive Context Windows
Feb 28, 2026
- EP066: Llama 2 Ghost Attention And Safety Secrets
Feb 28, 2026
- EP065: Teaching Small AI To Think Like Giants
Feb 28, 2026
- EP064: Synthetic Textbooks Break AI Scaling Laws
Feb 28, 2026
- EP063: RWKV Smashes the Transformer Memory Ceiling
Feb 28, 2026
- EP062: VOYAGER AI Masters Minecraft by Writing Code
Feb 28, 2026
- EP061: Fine-Tuning LLaMA 65B on One GPU
Feb 28, 2026
- EP060: Direct Preference Optimization Replaces RLHF
Feb 27, 2026
- EP059: Tree of Thoughts Unlocks System 2 Thinking
Feb 27, 2026
- EP058: Inside the Autonomous AI Town of Smallville
Feb 27, 2026
- EP057: Blind GPT-4 Taught LLaVA To See
Feb 27, 2026
- EP056: Pythia Turns AI Alchemy Into Chemistry
Feb 27, 2026
- EP055: Can GPT-4 Fairly Judge Other AI
Feb 27, 2026
- EP054: Alpaca - Stanford Built a $600 GPT Clone
Feb 27, 2026
- EP053: Sparks of AGI in Early GPT-4
Feb 27, 2026
- EP052: GPT-4 Bar Exam and Visual Reasoning
Feb 27, 2026
- EP051: ControlNet Solves Spatial Control With Zero Convolutions
Feb 27, 2026
- EP050: How Meta's LLaMA Beat GPT-3
Feb 27, 2026
- EP049: Toolformer Teaches Itself to Use APIs
Feb 27, 2026
- EP048: BLIP-2 Teaches Frozen Models to See
Feb 27, 2026
- EP047: Bootstrapping AI With Self-Generated Instructions
Feb 27, 2026
- EP046: Training AI With A Constitution
Feb 27, 2026
- EP045: BLOOM The Open Source Rival To GPT-3
Feb 27, 2026
- EP044: How ReAct Synergizes Reasoning and Acting
Feb 27, 2026
- EP043: Weak Supervision Made OpenAI Whisper Robust
Feb 27, 2026
- EP042: Running 175B Models on Consumer Hardware
Feb 27, 2026
- EP041: FlashAttention Smashes the AI Memory Wall
Feb 27, 2026
- EP040: Meta's Open Source GPT-3 Replica
Feb 26, 2026
- EP039: Flamingo Unlocks Few-Shot Visual Reasoning
Feb 26, 2026
- EP038: PaLM's 540 Billion Parameters Unlock Reasoning
Feb 26, 2026
- EP037: DeepMind Chinchilla Ends The Parameter Wars
Feb 26, 2026
- EP036: How 40 People Taught GPT-3 Manners
Feb 26, 2026
- EP035: How Google LaMDA Learned To Use Tools
Feb 26, 2026
- EP034: Chain of Thought Prompting Unlocks Reasoning
Feb 26, 2026
- EP033: Democratizing Image Generation with Latent Diffusion
Feb 26, 2026
- EP032: WebGPT Fights Hallucinations With Web Search
Feb 26, 2026
- EP031: DeepMind RETRO Swaps Memorization For Retrieval
Feb 26, 2026
- EP030: DeepMind's Gopher Exposes Limits of Scale
Feb 26, 2026
- EP029: Instruction Tuning Unlocked Zero-Shot Learning
Feb 26, 2026
- EP028: Train Short for Infinite Context
Feb 26, 2026
- EP027: From Creative Writer to Logic Engine
Feb 26, 2026
- EP026: LoRA Fine-Tunes Massive Models Without Supercomputers
Feb 26, 2026
- EP025: RoPE Solves Sequence by Rotating Vectors
Feb 26, 2026
- EP024: OpenAI CLIP Bridges Language and Vision
Feb 26, 2026
- EP023: Scaling Switch Transformers to Trillion Parameters
Feb 26, 2026
- EP022: DALL-E Treats Images Like Language
Feb 25, 2026
- EP021: Vision Transformers Beat CNNs at Scale
Feb 25, 2026
- EP020: Big Bird Scales Transformers With Sparse Attention
Feb 25, 2026
- EP019: Facebook's Linformer Solves the Attention Bottleneck
Feb 25, 2026
- EP018: Turning Digital Static Into Images With Diffusion
Feb 25, 2026
- EP017: RAG Gives AI a Library Card
Feb 25, 2026
- EP016: GPT-3 Learns From Examples Without Retraining
Feb 25, 2026
- EP015: Longformer Smashes the 512 Token Barrier
Feb 25, 2026
- EP014: ELECTRA Beats GPT On One GPU
Feb 25, 2026
- EP013: Reformer Cracked the Transformer Memory Wall
Feb 25, 2026
- EP012: Google T5 Turns Every Task Into Text
Feb 25, 2026
- EP011: ZeRO Solved the Trillion Parameter Memory Wall
Feb 24, 2026
- EP010: ALBERT Outperforms BERT With Parameter Sharing
Feb 24, 2026
- EP009: Slicing the AI Brain with Megatron-LM
Feb 24, 2026
- EP008: RoBERTa Proves BERT Was Just Undertrained
Feb 24, 2026
- EP007: How GPT-2 Hallucinated Ovid's Unicorn
Feb 24, 2026
- EP006: Transformer-XL Cures AI Amnesia
Feb 24, 2026
- EP005: How BERT Mastered Language by Hiding Words
Feb 24, 2026
- EP004: How 7000 Unpublished Books Birthed GPT
Feb 23, 2026
- EP003: How ELMo Made Word Vectors Dynamic
Feb 23, 2026
- EP002: ULMFiT Was the ImageNet Moment for Text
Feb 23, 2026
- EP001: How Transformers Smashed the Sequential Bottleneck
Feb 22, 2026