4 Ways to Make Sure Your Agency Can Trust Your AI Tools

• October 1, 2026

By Doug Marquis

Buying artificial intelligence (AI) is easy. Getting people to trust and adopt it is much harder.

Employees are nervous about learning something new or making a mistake that might embarrass them or damage their reputation, while executives are unsure of the right safeguards to put in place. Until agencies address those concerns, AI adoption will remain slow and uneven.

Building trust is less about the technology than the policies, training and oversight behind it. Here are four steps agencies can take to make AI a tool employees will use:

1) Build confidence through training and small, visible wins. Despite all the attention AI receives, many employees still aren’t comfortable using it and are reluctant to admit it. In addition to providing “how-to” training, agency leaders can remind employees that agencies already trust it more than they realize. For example, claims and underwriting have used machine learning for years.

Start with everyday tasks where the results are easy to verify, such as drafting emails, summarizing documents or brainstorming ideas. Short lunch-and-learn sessions showing real, small, visible wins build confidence in the technology, making agencies more comfortable expanding the technology into other parts of their work.

2) Use a layered approach to governance. People sometimes make the mistake of creating a blanket AI policy that governs everything. In reality, that puts handcuffs on employees’ ability to do things. Not every use of AI carries the same level of risk, so your governance and oversight shouldn’t be identical, either. Internal brainstorming with AI requires different safeguards than client-facing communications or applications that process personally identifiable information (PII). Match your policies to the level of risk rather than applying the same rules to every use case.

Good governance isn’t a thick rulebook that covers every possible scenario. It’s understanding how AI tools work, knowing how your data is used and protected, and giving employees clear guidance about what they can and cannot do.

3) Adjust human oversight as AI proves itself. Early on, every AI-generated recommendation should be reviewed by a person. Think of that oversight as a dial, not a switch. As confidence grows in AI through testing and monitoring, the dial turns down gradually rather than flipping off altogether. A manager might move from reviewing every recommendation to auditing a sample each week. No matter how far the dial turns, the human stays accountable for whatever the AI produces.

4) Treat trust as an ongoing process. Trust in AI isn’t something an agency earns once and then puts aside. As AI tools change and use cases evolve, agencies need to continually evaluate how AI is being used, ensuring it remains useful, safe and aligned with the agency’s goals. Also, remember that AI systems change over time as models are updated and new data is introduced. Even mature AI systems should be monitored regularly because people, not AI, remain accountable for the decisions they make.

The agencies that succeed with AI understand where AI can help, where human judgment is essential and how to safeguard both. Do that well, and the real payoff isn’t speed. Instead, it’s what that time buys back. That is the true return of AI adoption.

Doug Marquis is chief technology officer of Zywave, a Milwaukee-based InsurTech that delivers unmatched insurance intelligence using the Zywave AI Apex platform so insurance brokers, agencies, MGAs and insurance carriers can grow their business.