Million-Token Context Still Fails the Workload Test
Anthropic reported on February 5, 2026 that Claude Opus 4.6 scored 76% on the 8-needle 1M-token MRCR v2 test while Claude Sonnet 4.5 scored 18.5% on the...
Swarm Signal - AI Research for People Who Build
Technical AI research, explained clearly for researchers, builders, and anyone trying to understand what actually matters.
Agent Evals Need Harnesses, Not More Scoreboards
AgentCompass first landed on arXiv on 15 July 2026 with a practical complaint: agent evaluation is fragmented, tightly coupled, and hard to reproduce...
Coding Agent Benchmarks Hit the Generalization Wall
Scale's SWE-Bench Pro public leaderboard reports that top models scoring above 70% on SWE-Bench Verified fall to 23.3% for OpenAI GPT-5 and 23.1% for...
Agent Test-Time Scaling Needs Reuse, Not More Rollouts
General AgentBench reports that running agents for more interaction steps or more sampled trajectories did not reliably improve ten leading agents,...
The Goldfish Brain Problem: Why AI Agents Forget and How to Fix It
▶️ LISTEN TO THIS ARTICLE Your browser does not support the audio element. The Goldfish Brain Problem: Why AI Agents Forget and How to Fix It By Tyler Casey · AI-assisted research & drafting · Human editorial oversight @getboski In April 2023, a Stanford research team deployed 25 generative agents into a simulated
Multi-Agent Finance Workflows Need Cost Curves, Not More Agents
A March 2026 benchmark on financial-document processing makes the uncomfortable point: the most accurate multi-agent architecture was not the obvious...
Self-Improving Agents Have an Evaluator Problem
Anthropic's June 2026 update on recursive self-improvement is not a distant sci-fi warning. The company says its engineers now ship 8x as much code per...
Coding Agents Need Trajectory Reviews, Not Pass Bits
Most coding-agent benchmarks still compress a whole run into one bit: did the task pass? AgentLens argues that users experience the whole trajectory...
The Agent Project That Should Have Been One LLM Call
Some enterprise agent projects fail because autonomy was added where a bounded single-call LLM design would have delivered cleaner behavior and lower operational risk.
Agent Marketplaces Need Abuse Screens, Not Escrow
Agent marketplaces are no longer only payment demos; they are becoming tool surfaces where software can hire people. A February 2026 empirical study of...
Practical tools
Templates for budgets, projects and freelance work
BoredTools offers practical spreadsheets for planning work, tracking money and running small projects.