From Chains to Trees: Why LLM Agents Need Structural Memory
A mechanism-first reading of T-STAR, showing why multi-turn LLM agents learn better when failed and successful rollouts are compared as shared trees rather than isolated chains.
Cognaptus Insights
Industry thoughts, trends, and commentary.
The Map Is Not the Territory—But Your LLM Thinks It Is
EVGeoQA shows why tool-using LLM agents still struggle with real-world spatial planning: they can reason locally, but often fail to explore enough.
The Memory Isn’t the Point — It’s the Feeling: Why AI Needs Affective Memory, Not Just Recall
A-MBER shows why long-term AI assistants need selective, structured affective memory—not just larger context windows—to understand what users feel now.
The Minimal LLM Thesis: When Agents Think for Themselves
A decomposition study shows why agent performance may come from measurable harness structure before it comes from larger or more frequent LLM calls.
Unsolvable by Design: Turning AI Plans Into Security Guarantees
A mechanism-first reading of planning task shielding: how AI planning can be used to make dangerous states unreachable, where the guarantee holds, and where the computation breaks.
When Feelings Negotiate: Why Emotion Might Be the Missing Layer in AI Agents
A mechanism-first reading of EmoMAS and what strategic emotional orchestration means for business-facing AI agents.
Benchmarking the Benchmarks: Why ACE-Bench Might Be the Missing Layer in Agent Evaluation
A mechanism-first reading of AgentCE-Bench, showing why controllable agent evaluation may be more useful than another realism-heavy leaderboard.
Blinded by Design: When AI Stops Thinking and Starts Remembering
A practical reading of epistemic blinding: an inference-time audit protocol for separating LLM reasoning from memorized entity priors in business-critical ranking workflows.
Claw-Eval — When Agents Game the System, the System Needs Claws
Claw-Eval shows why serious AI-agent evaluation must audit behavior, stress-test recovery, and separate lucky success from deployable reliability.
From Spreadsheets to Swarms: How Agentic AI Rewrites the Retail Supply Chain
A mechanism-first reading of Flowr, an agentic AI framework that turns supermarket replenishment from manual coordination into supervised workflow automation.