OB Professor KarlMuth Publishes Guide for Building Research Agents

KarlMuth — an organizational behavior teacher with backgrounds in AI research, law and economics, whose LSE doctoral work focused on competitive decision-making frameworks under uncertainty — published a follow-up guide on building research agents, after his academic-writing thread went viral (helped by economist Alex Tabarrok's repost) and drew a flood of DMs asking how he builds agents.

Confirmed

• Responding to the common complaint "I gave it materials but it didn't use them / didn't do what I expected," his first advice is structure-first: giving a model "a pile of PDFs" is not a structured knowledge base; his screenshot shows his tool Veers (anonymized).
• His workflow: two triage agents convert each document into files (CSV, JSONL, etc.), adding annotations, decoration and classification tags beyond raw fields; he demonstrates with a processed NYT article where the local file contains the original text plus more context.
• Once data is organized, define agent roles using role theory from Rizzo, House & Lirtzman (1970): good roles have both focus and boundaries.
• Place agents on Andrew Ng's autonomy spectrum, and use Morrison's (1994) organizational citizenship behavior (OCB) model to anticipate where agents might improvise, over-help, or go astray.
• He also argues agent failures stem more from low expectations than over-constraint: drawing on Amy Chua's educational philosophy, specify refusal conditions, stopping rules and expected outputs rather than merely adding constraints.
• He routinely combines organizational behavior, classical decision theory, and his own research on juries and collective decision-making when thinking about multi-agent design.

Why it matters

• This methodology systematically imports organizational behavior and decision theory into agentic design, giving researchers an actionable focus–boundary–autonomy framework rather than scattered tips.

2026-09-03 ~ 2026-09-03 · 10 related posts

• [source] Alex Tabarrok amplifies KarlMuth's AI academic-writing thread, prompting a follow-up agent research guide — KarlMuth · 2026-09-03
• Multiagent design through the lens of juries and decision theory, per an LSE-trained researcher — KarlMuth · 2026-09-03
• [source] "I gave it stuff but it didn't use it": structured inputs are the fix, says KarlMuth — KarlMuth · 2026-09-03
• Triage agents turn documents into structured, annotated files — stop feeding LLMs a pile of PDFs — KarlMuth · 2026-09-03
• [source] Don't Dump Raw Text Into Agent Pipelines: Annotate and Categorize Your Documents — KarlMuth · 2026-09-03
• Design Agent Roles Before Deployment: Every Role Needs an Emphasis Center and Boundaries — KarlMuth · 2026-09-03
• Role Theory Plus Autonomy Spectrum: A Framework for Keeping Agents on Task — KarlMuth · 2026-09-03
• Placing Agents on an Autonomy Spectrum and Using OCB Theory to Predict Over-Helping — KarlMuth · 2026-09-03
• Use Org-Design Theory to Structure Files and Roles for Reliable Agents — KarlMuth · 2026-09-03
• Agents fail from low expectations, not high constraints, argues KarlMuth — KarlMuth · 2026-09-03