AI Agents

TL;DR for operators AIM-Bench is not another “which model is smartest?” leaderboard. It is a warehouse stress test for agentic LLMs asked to make replenishment decisions under uncertainty.1 The useful lesson is uncomfortable: inventory agents can look mathematically fluent while still behaving like biased managers. Most evaluated models show mean anchoring in the newsvendor task. All evaluated models show bullwhip amplification in the Beer Game. Some models over-order to avoid stockouts; others keep leaner inventory but accept higher shortage risk. In other words, the operational personality of the model matters. ...

TL;DR for operators Portfolio teams do not need another chatbot that confidently explains why yesterday’s price move was “driven by sentiment.” They need a system that can split research work into specialised roles, force disagreement into the open, log the reasoning trail, and turn messy inputs into a decision that a human can inspect before money moves. ...

TL;DR for operators Most agent teams eventually discover that “the workflow” is not one thing. A customer-support agent, a coding agent, and a mathematical reasoning agent may all use decomposition, verification, consensus, and answer extraction—but not in the same order, not with the same emphasis, and definitely not with the same failure modes. Static agent templates look tidy in architecture diagrams. Then the first heterogeneous workload arrives, and the diagram starts quietly sweating. ...

TL;DR for operators Most enterprise RAG failures are not generation failures. They are retrieval-routing failures wearing a very convincing blazer. The paper behind this article proposes an open-source agentic hybrid RAG framework for scientific literature review: bibliographic metadata and citation relationships go into a Neo4j knowledge graph; full-text PDF chunks go into a FAISS vector store; an LLM-based agent decides whether a user’s question should be answered through GraphRAG or VectorRAG; a Mistral-based generator produces the final answer; DPO is used to improve grounding; and bootstrap resampling is used to report evaluation uncertainty.1 ...

TL;DR for operators A new survey on agent workflows is not useful because it tells us agents are becoming important. Anyone still surprised by that has probably been trapped in a quarterly innovation committee. Its value is more practical: it turns the messy agent-tool-platform landscape into a comparison map for deciding what kind of workflow infrastructure a business is actually buying or building.1 ...

TL;DR for operators A user asks an AI agent to delete an account. The obvious tool is DeleteAccount. A normal semantic retriever will probably find it. Splendid. The agent still fails if it misses GetUserToken, because the deletion tool needs a token first. This is the failure mode Tool Graph Retriever, or TGR, is built to address.1 ...

TL;DR for operators The useful reading of this paper is not “AI agents are coming to crypto.” That is already obvious, and in some corners of the market, painfully over-branded. The sharper point is that Web3-AI agents are forming a stack. At the bottom are infrastructure and trust layers: protocols, DePIN systems, verification mechanisms, execution environments, and agent-development platforms. On top sit the applications: DeFi agents, portfolio tools, market-intelligence systems, governance assistants, security auditors, creative agents, and RWA managers. The paper’s dataset of 133 projects shows this stack is not evenly valued. Infrastructure accounts for 67.8% of the analysed $6.92 billion market capitalisation, even though incubation platforms show the most project activity.1 ...

TL;DR for operators AI shopping agents do not simply “find the best product.” They convert a messy human browsing process into a model-mediated allocation system. That allocation system has its own priors, positional quirks, trust cues, and semantic blind spots. Lovely. We automated the customer and discovered a new customer. The paper introduces ACES, a controlled sandbox for testing AI shopping behaviour. It pairs a browser-use or API-style buying agent with a programmable mock e-commerce site, then randomises product order, prices, ratings, reviews, badges, and product descriptions to estimate what actually moves an AI agent’s choice.1 ...

TL;DR for operators A familiar enterprise AI failure looks like this: the model gives a confident answer, the formatting is exquisite, the explanation sounds like a gifted teaching assistant, and one equation quietly takes the project into a ditch. Physics is an unusually good place to study that failure because being clear is not enough. The system must interpret the situation, select the right principle, keep the units straight, calculate correctly, and not hallucinate a helpful-but-illegal assumption because the prompt looked lonely. ...

TL;DR for operators Most financial systems are designed as if finance begins with institutions: contracts, lenders, insurers, markets, prices, and enforcement. Paper 2506.00099 asks a cleaner question: what if the core behaviours behind finance emerge before those institutions, from repeated reciprocal interaction?1 The paper’s central move is to treat trade as the simplest case of reciprocity, then derive credit, insurance, token exchange, and investment as structural extensions of the same mechanism. Add delay, and reciprocity starts to look like credit. Add asymmetric risk, and it starts to look like insurance. Add portable mediation, and it starts to look like token exchange. Add expected future reward, and it starts to look like investment. Finance, in this view, is not born fully dressed in a suit carrying a term sheet. It begins as remembered obligation. ...