Multi-step reliability

Where it is assessed

5 assessed items, 0 done.

• AIE-104 M1.5 Compounding error, on a system you own · tier 1
• AIE-104 M4.3 pass^k on your own agent · tier 1
• AIE-104 M4.4 Design against the decay · tier 1
• AIE-104 M4.5 Retention: reliability metrics and interventions · tier 1
• AIE-104 M9.1 Ship an agent that knows its own limits · tier 3

Verified sources

• How we built our multi-agent research system
• Agentic AI MOOC Fall 2025 · Berkeley RDI
• Reflexion: Language Agents with Verbal Reinforcement Learning
• tau-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains
• Why Do Multi-Agent LLM Systems Fail?