Buying Insurance for Myself Was Harder Than Writing Code

How It Started

Simple enough: I wanted to buy health insurance.

Seems totally ordinary, right? Except — my physical had flagged a pile of "minor issues." Nodules, hyperplasia, gastritis, cysts. Individually, none of them are a big deal. But in the insurance industry, they have a name for people like me: non-standard applicants.

Buying insurance as a non-standard applicant versus buying it as a healthy person are completely different experiences.

Just How Painful Is It

1. Finding Something You're Even Allowed to Buy Is Already a Battle

For healthy people, buying medical insurance is easy — open any platform, pick a popular product, submit. For non-standard applicants? First you have to get through the health disclosure questionnaire.

Most million-yuan medical plans have things like "thyroid nodules" and "breast nodules" written right into the health disclosure. The moment you see them, you know — this one's off the table.

So step one isn't "which is the best option" — it's "which one will actually let me in."

I eventually discovered a category of product called health-disclosure-free medical insurance — no health questions asked, buy straight in. Sounds great:

Then I read the fine print, and it was a completely different story.

2. "No Health Disclosure" Doesn't Mean "Everything Is Covered"

• Pre-existing conditions are split into "general" and "major" categories
• General ones may be covered; major ones (cancer, cirrhosis, coronary heart disease, etc. — 5 categories) are not
• Some products are "insurable and claimable," others are "insurable but not claimable," and still others say "no payout for the first 3 years, covered from year 4"
• Every insurer defines these terms differently

Just getting clear on the distinctions between these concepts ate up an entire evening.

3. The Policy Terms Are a Wall

Pull up the coverage disclosure for any product and it looks like this:

Every line uses legal language to describe a limitation you'd probably never notice on your own. What's excluded, which waiting periods get calculated separately, which coverage needs to be added as a rider — you won't know what you actually bought unless you read every single word.

4. Pricing Is Hidden Where You'd Never Think to Look

Want to compare prices? Not that simple. Some products hide their premium tables inside JavaScript variables on the page (window.priceRules), others in static JS files (price_rate_v*.js), and some require you to fake a mobile browser just to intercept the data.

This rate table, for example, was pulled straight from the page source:

And here's another product's pricing structure — sliced by age group, with/without social insurance, and different plan tiers, resulting in dozens of price combinations:

7 products × a dozen dimensions each. Building this spreadsheet by hand would drive you insane.

5. Renewal Terms Are the Biggest Word Game of All

"Guaranteed renewable for 5 years" sounds reassuring — until you read carefully:

• You must pay renewal premium within 60 days of policy expiry; miss it and you permanently lose renewal rights
• If cumulative claims reach the cap within the 5-year period, coverage terminates immediately
• Renewal after the 5-year period requires fresh underwriting by the insurer; if the product is discontinued, you're out
• Premium rates can be adjusted — they may go up

Meanwhile, "non-guaranteed renewable" products mostly do renew in practice; the real risk is just product discontinuation. These distinctions are completely invisible unless you pull apart the policy wording clause by clause.

What AI Actually Did

Honestly, without AI, I would have probably given up and let this drag on indefinitely.

Breaking Down Policy Terms

Insurance policy PDFs run dozens of pages and are packed with legalese. Drop the PDF into AI, ask it to extract the key clauses, flag the critical restrictions, and translate everything into plain language. Done in minutes — something that would have taken me two or three hours to read myself.

Calculating Premiums

AI helped write scrapers, parse the rate tables, and calculate actual prices by age group for side-by-side comparison. Those figures buried in page source code would have been a nightmare to pull manually.

Structured Comparison

AI generated an interactive comparison table — organized by dimension, color-coded by quality, filterable by scenario:

The final output was 5 deep-analysis reports from different angles: product overview, premium analysis, non-standard applicant match assessment, renewal research, and multi-layer coverage combination strategies.

Surfacing Hidden Details

While analyzing the policy terms, AI caught something a human reader would almost certainly have missed: ZhongAn's Zunxiang e-Life mid-to-high-end plan has a "3-year pre-existing condition waiver" mechanism — after 3 consecutive years of enrollment, pre-existing conditions become claimable starting in year 4. This was buried deep in the policy. The product's marketing page mentioned none of it.

A Few Real Lessons

1. For non-standard applicants, the question isn't "which is cheapest" — it's "which will actually pay out." No-disclosure-required is just your entry ticket; how the insurer handles pre-existing conditions is what actually matters.

2. Don't put all your coverage in one product. If a non-guaranteed-renewable product gets discontinued, you — now in worse health than when you first applied — may struggle to find a replacement. The smart play is a guaranteed-renewable product as your foundation, supplemented by a no-disclosure product.

3. Every word in an insurance policy means something. "Insurable" is not the same as "claimable." "Guaranteed renewal" and "automatic renewal" are two different things. The way deductibles are calculated varies by insurer.

4. AI can't make the decision for you, but it can help you see the full picture. Fetching, parsing, structuring, and comparing information scattered across dozens of sources — these tasks are high-repetition but demand extreme attention to detail. AI does them better than humans. You just need to make the final call.

5. The information asymmetry in the insurance industry is real. Different agents at the same company may give you completely different takes on their own products. A PICC agent I spoke with didn't even recommend the product their company co-developed. Cross-validating across multiple sources is essential.

Closing Thought

Buying insurance is fundamentally an information-intensive decision. Too many products, complex terms, widely varying personal circumstances, information scattered everywhere. This is exactly where AI shines — not by "recommending" a product for you, but by laying out every option, translating every detail into plain language, and marking the hidden traps.

The final call is yours. But at least you're making it with a clear view of the whole picture.