What's actually on page 34 of your insurance policy?

Today's Tip: Reading Your Insurance Policy

You are paying for an insurance policy you have probably never read. And buried inside it are exclusions that could cost you thousands of dollars when you need coverage most.

Let me explain…

Insurance policies—homeowner’s, auto, all of them—are written in dense language. Phrases like “concurrent causation” and “ordinance or law coverage” are designed for lawyers, not for people like you and me.

So most people do what feels reasonable… they file the policy in a drawer, pay the premium every six months, and hope for the best. Sound familiar?

My friend Gloria found out the hard way what that habit can cost. Gloria is 64, a retired school librarian from Raleigh, North Carolina. Last spring, a big storm knocked a tree limb through her back porch roof. Rainwater poured in and ruined the porch furniture, the decking, and part of the drywall inside her sunroom.

She called her insurance company, feeling confident. “I’ve been paying this policy for 19 years,” she told me. “Of course it’s covered.”

Except… it wasn’t. Not entirely. The adjuster told her the tree removal was covered, and the roof repair was covered.

But the porch furniture? Not covered—her policy had a specific exclusion for “outdoor personal property damaged by wind or rain.” And the interior water damage? Subject to a separate, higher deductible she didn’t know she had.

Gloria was out about $4,200 she thought was covered. It was sitting right there in her policy… on page 34. She just never knew to look.

Here’s the incredible part. After that experience, Gloria’s daughter suggested she try uploading the policy to Claude—the free AI tool from Anthropic. Gloria was skeptical. But she gave it a shot. She uploaded the 47-page PDF, typed a single prompt… and in about 90 seconds, she had a plain-English summary of exactly what was covered, what was excluded, and what her deductibles were for each type of claim.

“I felt like an idiot,” Gloria said. “Not because I couldn’t read it before—because nobody should have to.”

She’s right. And you can do the same thing right now… with any policy you have. Homeowner’s, auto, umbrella, renters—it works on all of them. Here is the exact prompt Gloria used.

When Gloria ran this prompt, she discovered three things in ten minutes that she’d missed in 19 years. First… her policy excluded sewer backup damage unless she paid for an add-on rider. She had no rider. Second… her personal property coverage was capped at $1,000 per item, meaning her piano and her late husband’s coin collection were wildly underinsured. Third… she had a 72-hour reporting window for theft claims that she’d never heard of.

Three blind spots. All fixable… once she knew they existed. She called her agent the next day, added the sewer backup rider for $47 a year, and scheduled a policy review to increase her personal property limits.

That is the power here. Not that AI replaces your insurance agent… but that it prepares you to have a much better conversation with your agent. You show up knowing the right questions instead of nodding along.

Now, I know what you’re thinking… “Can I really trust AI to read a legal document correctly? What if it misses something important?”

Fair question. And the honest answer is… AI can misread a clause, especially if the policy uses unusual formatting or references another document. That’s why this is a starting point, not a finish line.

You use AI to find the sections that matter most to you… then you read those specific pages yourself, or you bring them to your agent and say, “Walk me through this part.” That is a completely different conversation than “So… what does my policy cover?”

Think of it this way. Before AI, you had two options: read pages of dense legalese yourself… or trust that everything was fine. Now you have a third option—get a guided tour of your own policy in plain English, then verify the parts that matter most.

Here’s what to do next. Find your homeowner’s or auto insurance policy—check your email, your filing cabinet, or your insurance company’s website. Most carriers let you download a PDF from your online account.

Upload it to Claude at claude.ai. Paste the prompt above. Read what comes back.

If everything looks solid… great, you have peace of mind you didn’t have before. If something surprises you… even better. Now you know about it before a storm, a break-in, or a fender bender forces you to find out the hard way.

Gloria told me she wishes she’d done this years ago. You can do it today… in about five minutes.

★ Quick Wins

Skill Builder: Getting AI to Cite Its Sources

Here’s a problem you’ve probably already noticed… AI gives you an answer, and it sounds confident. Very confident. But where did that information come from? Is it pulling from a government website, a random blog post, or something it made up entirely?

By default, most AI tools don’t tell you. They give you the answer and expect you to trust it. A recent study found that popular AI models gave wrong financial advice 57% of the time. That’s worse than a coin flip. So trusting without checking is a risk you don’t need to take.

The fix is surprisingly easy… you ask the AI to show its work. When you add one line to your prompt—“cite your sources”—the AI shifts from giving you a confident opinion to giving you a supported argument. It will name specific websites, organizations, or documents. And that gives you something you can actually verify.

But here’s the important part… “cite your sources” alone isn’t specific enough. You want to tell the AI what kind of sources you trust. Government sites? Medical journals? The company’s own website?

The more specific you are about where you want the information to come from, the more useful the citations become. Here’s a prompt that does this well.

One warning… even when AI cites sources, it can sometimes generate fake URLs or attribute information to the wrong organization. Always click the links it gives you and make sure they lead where they claim. If a link is broken or the page doesn’t say what the AI claimed it said… that’s your signal to dig deeper or ask a professional.

The goal isn’t to make AI perfect. The goal is to make it accountable. When you ask for sources, you change the relationship… from “I hope this is right” to “I can check this myself.” And that is a massive difference.