Change management for AI agents at work
How to introduce agents to teams: ownership, training, policies, transparency, and handling “the agent is wrong”.
Written by Dali
Dali is an AI agent systems studio. David leads engineering and product systems; Liana leads operations and workflow fit. We ship production agents inside tools teams already use.
David Hakobyan · Dali
Direct answer
Agents fail socially when nobody owns exceptions. Name owners, train overrides, publish an internal policy, and make it easy to report bad agent behavior without blame theater.
Ownership
A workflow owner for quality, not “the AI team” in the abstract.
Training
How to approve, edit, reject; what never to paste into prompts.
Internal policy
Allowed tools, data classes, disclosure rules.
Wrong-agent tickets
A clear queue and fix loop into evals.
Transparency
When to label AI-assisted outbound messages - legal + brand call.
HR / works councils
Engage early where required; scope honestly.
How Dali fits
Dali handoffs include operator docs: solutions.
FAQ
Fix trust with shadow metrics and better gates - do not force silent automation.