Must-read paper on the effectiveness of multi-agent systems and persistent memory. We have recently seen many viral examples of multi-agent systems scaled via persistent memory. This work takes a closer look. They test one three-tier chain-of-agents system. Splitting long-context work across agents cuts peak KV cache to 14.3 MiB per query, against about 35 MiB for single-pass and RAG baselines. The persistent trace-recall tier adds 0.368 MiB of peak cache and changes accuracy by +0.015, with a confidence interval that crosses zero. The authors explain why. Single-question benchmarks give each item its own evidence and score it independently, so stored traces have nothing useful to recall. Paper: Chat with Paper:

Must-read report by Terawatt PV Research on US Solar market. $CSIQ in second place. $FSLR ($20B mkt cap) is first place. $GLW (+$100B mkt cap) is tied in second with CSIQ. $TE , $TOYO , QCells, SEG, Illuminate, etc… all below $CSIQ , as they should. This is one of the first (high-quality) research reports that gives Canadian Solar the credit they deserve. Behind FirstSolar, they are the leading US manufacturer of solar modules. And the lead is widening. +$4.5B & 13GWs in CONTRACTED backlog. Customers are pre-paying for their orders and helping fund their mfg expansion. Neither customers, nor lawyers, nor consultants, nor tax equity financiers are seeing FEOC risks. Anyone who thinks $TE has some kind of lead, has clearly not studied this market enough… The slow bleeding has been extremely hard, but it will turn. The value embedded in this stock is MULTIPLES of their market cap.

Must-read paper from Google on self-improving agent harnesses. If you auto-optimize your agent's harness, your eval score can go up while the agent gets worse on real tasks. This paper shows how to prevent that. Of five harness-evolution methods compared on agentic workspace tasks, RRSI scored the lowest on the tasks it evolved against and highest on all three out-of-distribution benchmarks. Automated harness evolution proposes edits to prompts, control flow, tools and memory, keeps the ones that raise the score, and repeats. The authors show this overfits the training tasks. Meta-Harness reached 93.0 on the Harvey LAB evolve split but gained only 0.3 to 1.5 points on JobBench, GDPval and APEX-Agents. RRSI adds regularization on both sides of the loop. The proposer gets an edit budget that shrinks over time and is pushed toward directions it has not tried. A critic rejects benchmark-specific edits, and a pruner removes edits that are too small, too costly or no longer useful. RRSI scored 90.5 on the evolve split and gained 3.5 to 4.7 points on the three held-out benchmarks. In the ablation, unregularized evolution used 3.80M tokens per trial against 2.42M for RRSI. With Gemini 3.5 Flash, RRSI raised Terminal-Bench 2.1 from 64.6 to 78.7 and carried a 2.2-point gain over to SWE-bench Verified. Paper: Chat with Paper:

Must-read papers of the week ▪️ Agent Lightning v1.0: Towards Harnessed Agentic RL ▪️ LEGO-RL: Harness-Native Reinforcement Learning for Coding Agents ▪️ EnvHarness: Awakening Static Worlds for Agent Learning ▪️ SPADE: Self-Play in Adaptive Synthetic Executable Environments ▪️ SkillGate: Training In-Policy Skill Selection in Long-Horizon Agents ▪️ Looped Language Models Improve Compositional Tool Calling ▪️ Graph Engineering in the Era of LLM Agents ▪️ Chain-of-Experience for Continual LLM Improvement ▪️ Cross-Model Memory Transfer via Target-Side Reader Adaptation ▪️ CLEAR: Continuous Latent Adapter Routing for Utility-Preserving LLM Safety Alignment ▪️ EXIMO: VLM Guided Exploration of VLA Policies ▪️ FreeToken: Efficient Edge-Native MoE Serving with Bandwidth-Adaptive Execution ▪️ Llama-Mobile: Efficient 2.7-Bit Quantization of VLMs ▪️ Improving the Matrix Multiplication Exponent with Modern Optimization and AlphaEvolve Here is the full list of the most interesting papers + links:

Must-read papers of the week ▪️ AI4AI at Test-Time: Strong-to-Weak Capability Transfer via Harnesses ▪️ Persistent Recursive Worlds Enable Autonomous Software Evolution ▪️ AgentRewind: Recoverable Execution for Long-Horizon LLM Agents ▪️ Demystifying Agent Skills: Why They Work-Until They Don’t ▪️ Envs-FORGE: Frontier-Optimized Reward-Grounded Environment Synthesis for Agent RL ▪️ Stealing Reasoning Traces from Proprietary LLM APIs ▪️ Agent Safety Should Be a Runtime Contract ▪️ WorldClaw: Agentic 3D Open-World Generation at Scale ▪️ Second Thought: Reasoning in Parallel as LLM Agents Act and Observe ▪️ ScienceFlow: A Long-Horizon Agent for ML Research, Scientific Discovery and Beyond ▪️ OmniScientist: An Omni-Modal Omni-Discipline AI Scientist Find the full list of the most interesting papers + links here:

MUST READ: Can inflation do what the bombs didn't? "There are days when I don’t have any money'" — Iranians struggle to afford basic necessities as Trump bets that intensifying economic pressure will succeed where military force has not. @golnarM

MUST READ from Lorissa Rinehart: "The Trump DOJ alleged case against Siebel Newsom lacks a single iota of validity…Since California’s First Partner is one of the most outspoken, dedicated, & effective proponents of women’s rights, MAGA has targeted her”