I'm building a multi-agent accountability system for AI development work and need your critical review.
I'm working with multiple AI agents (OpenClaw, Claude Code instances) on a product called Amplified Partners. We've had issues with commitments not being kept - not intentionally, but due to AI limitations (no persistence between sessions, context loss, etc.).