Back to Markets
Stocks● Neutral
Insurance AI Isn't a Tech Problem – It's a Design Problem
A design agency CEO argues the industry's AI investments fail because they don't start with the people using them. The solution is workflow integration, not automation.
Continue with
The CEO of Cake & Arrow, a design agency with more than 15 years in insurance, has a message for the industry: your AI investments aren't failing because of the models. They're failing because of the design.
His firm watches the same pattern repeat. An executive team picks a platform, builds a copilot, and rolls it out. Adoption stays low. The blame lands on the users. "Resistant to change" becomes the diagnosis. The CEO calls that wrong. People aren't resistant to tools that help them. They're resistant to tools that make their day harder, add complexity, or are designed to replace them.
The argument matters beyond the insurance conference circuit. Publicly traded insurtech companies – Lemonade, Root, Hippo, and others – have spent heavily on AI. Their stock performance has been mixed. If the CEO is right, the winners will be the ones solving real workflow problems, not the ones churning out automation features.
The Workflow Problem
Agents aren't asking for another tab, login, or disconnected assistant. They're already juggling agency management systems, CRMs, email, spreadsheets, carrier portals, and rating tools. They type the same information into multiple screens. They hunt for context across half a dozen apps.
The CEO calls this a workflow problem, not a single-task productivity problem. AI that writes an email is useful. AI that understands the context behind that email, pulls from the right systems, shows its sources, and lets the human verify – that's something different. That's connective tissue.
His firm spoke directly with agents and brokers for a recent report. What emerged was not resistance. It was resourcefulness. Agents are already experimenting with AI on their own. They draft emails, summarize policies, compare quotes, and pre-read meetings. Some build quiet workarounds because the official tools don't fit how they actually work.