What is resistance to change?

Resistance to change is the pushback people show when a new way of working is introduced, ranging from open objection to quiet workarounds and slow adoption. With AI agents, it often looks like reviewers redoing the agent's work by hand, refusing to approve anything, or approving everything to avoid owning the outcome.

Why does resistance to change matter with agents?

Resistance is information. Staff who push back on an agent are often reacting to a real gap: they do not know what they are accountable for, they have seen the agent make a mistake, or nobody has explained why it holds some actions and not others. Treating all resistance as fear of job loss misses the signal.

It also distorts the control. A reviewer who distrusts the agent and redoes every draft makes the deployment slower and teaches nobody anything. A reviewer who has given up and approves everything is worse: the record shows a named approval on decisions nobody checked.

What does it usually signal?

The most common causes in agent deployments are practical:

• Unclear accountability: "If I approve it and it is wrong, is that on me?" This is the big one, and it is covered in staff resistance is usually about accountability, not jobs.
• No visibility of reasoning: people cannot see why the agent reached a verdict, so they cannot judge it.
• Bad first impressions: an early error that nobody explained.
• Loss of craft: experienced staff feel their judgement is being replaced rather than used.

Each has a different fix. Accountability needs a clear statement of what the reviewer's approval means. Visibility needs training on reading the verdict and the bound that decided it.

An example

A claims team pushes back on an agent that drafts settlement letters. Approvals stall; senior handlers rewrite every draft. A team lead asks why, and hears one answer repeatedly: "I don't know what I'm signing for." The firm runs a session on the decision chain, showing that each verdict names the bound that decided it and that a held letter waits for a named person who can see the sources. It also agrees that reviewers are accountable for checking the flagged source, not for re-deriving the whole claim. Within a month, rewrites fall and send-backs come with useful notes.

How is resistance to change different from change management?

Resistance is the response. Change management is the planned work of leading people through a change: communication, training, involvement, support. Good change management expects resistance and uses it to find what is missing.

Neither should be confused with change control, which is the technical process for approving and recording changes to an agent's brief, sources or limits. Change control governs the agent; change management deals with the people.

Which terms are related?

• Behaviour change in training: whether training actually alters what people do; see what behaviour change means in training.
• Change management: the discipline that plans for resistance.
• Adoption curve: how uptake spreads through a team over time.

For the broader approach, read training teams for AI agents.

Where the Academy fits

Much resistance eases when people understand what they are approving. Pryme Intelligence Academy's Foundations track covers why agents must be governed and how the decision chain works, and the Operator track covers human review in detail. Reviewers who know which bound decided a verdict can judge it rather than fear it.