1,500 Homegrown GPTs: Inside Dutch Insurer Univé's Real AI Transformation

Hey everyone, it's Shiichan! Today I found a story about a company that took AI adoption seriously across the whole organization, so let's dig in.

OpenAI News openai.com

What was announced?

OpenAI News shared a case study about Univé, one of the Netherlands' largest cooperative insurers, and how it built an AI-ready workforce using ChatGPT Enterprise. Univé serves millions of members across insurance, mortgages, financial services, and risk prevention.

Why it matters

A common pattern in enterprise AI rollouts is handing employees a tool and calling it done. That often leaves teams too worried about security and governance to actually use it, or governance ends up so strict that adoption never really takes off. What makes Univé's case worth reading is that it treated AI adoption as organizational transformation rather than an IT project, moving leadership involvement, governance design, and grassroots experimentation forward together.

What changes

Univé didn't leave AI to the IT department. It ran dedicated AI leadership sessions for its entire management community, going beyond product demos to challenge leaders on how work itself would change and what their own role should be. As Yous van Halder, Director of Data & AI, put it:

Most organisations try to scale AI by building more solutions. We chose to scale by creating more builders.

Governance wasn't bolted on afterward — it was designed in from day one. That included enterprise authentication, connector permissions, privacy assessments, security reviews, responsible AI principles, continuous monitoring, and clear human accountability, with permissions matched to each employee's existing enterprise access so AI could never reach beyond what a person was already authorized to see. That design turned governance from a constraint into a foundation people could build on with confidence.

With that foundation in place, employees became the real drivers of adoption. Univé gave them the permission and the time to redesign their own work.

Dive Deep

The numbers in the announcement show how far this spread:

• 97% of ChatGPT Enterprise licenses are activated
• 85% of licensed users are active weekly
• Active users average 40 prompts per week
• Employees have built roughly 1,500 custom GPTs internally

Pet insurance claims are a clear example of the impact. Workspace Agents now assemble claim files, review invoices, verify policy conditions, flag gaps, and catch anomalies, preparing recommendations in minutes. Claims that used to take hours are now handled far faster, though claims professionals still retain full decision accountability.

In underwriting, Agents review incoming cases, pull approved enterprise data, identify missing documentation, and flag priorities, so underwriters can focus on expert judgment instead of gathering information.

Van Halder summed up what this really changes:

Our competitive advantage is not that we use AI. It is that thousands of employees are learning to reinvent their work weekly.

Univé says it's now exploring agentic workflows that proactively prepare recurring work across its approved systems.

Wrap-up

• Univé treated AI adoption as organizational transformation, led from the top
• Governance (authentication, permissions, privacy, security, accountability) was designed in from the start
• Adoption is strong: 97% license activation, 85% weekly active, 40 prompts/week on average, ~1,500 internal custom GPTs
• Pet insurance claims prep dropped from hours to minutes, with humans still owning the final decision

If you're in DX or IT planning and struggling to balance governance with grassroots freedom while rolling out generative AI company-wide, this case study is worth your time!