Brett Chereskin
← Back to writing
AI & WorkJune 5, 2026 · 5 min read

How I Used Claude to Fight a $600 Insurance Denial — and Actually Filed a Regulator Complaint

Share

A few months ago a visit to a specialist's office turned into the kind of administrative mess most people quietly eat. A claim got handled wrong, the coverage I thought I had didn't show up the way I expected, and I was suddenly on the hook for a bill that didn't match my understanding of my own plan.

To be clear: this post is not about my health. The medical part is boring and private, and it stays that way. This is about the bureaucracy that wraps around it — the denials, the Explanation-of-Benefits letters written in a language no human speaks, the phone trees, the "we'll have to escalate that." That part is universal. And it's exactly the kind of fight AI is quietly very good at.

If this ends the way I think it should, it saves me over $600. But the money isn't the most interesting part of the story.

Why everyone gives up

Here's the dirty secret of insurance disputes: the process is designed to be exhausting. Not maliciously, necessarily. But the friction is the feature. Every denied claim is a bet that you won't have the time, energy, or stomach to push back. Most of the time, that bet pays off.

I almost proved it right. I'm a persistent person, I run operations for a living, and even I looked at that stack of paperwork and felt the pull to just pay it and move on. Fighting it correctly — reading the policy, understanding my rights, building a paper trail, writing it all up in a way an institution has to respond to — felt like more hours than $600 was worth.

That's the trap. The cost of fighting is mostly your time, and the system knows your time is expensive.

Bureaucracy wins by making persistence expensive. AI makes persistence cheap. That single shift is enough to change who wins these fights.

What Claude actually did

I didn't ask Claude to "handle my insurance." I treated it the way I treat AI on everything — leverage on the parts that are slow and opaque, while I keep the judgment.

  • Translated the dense language. I pasted in the policy sections and the EOB and asked for plain English: what each line meant and where the gap was between what I was billed and what my plan said. The jargon stopped being a wall.
  • Built the timeline. Every call, letter, and date organized into a clean chronological record — who said what, when. The paper trail is the entire ballgame in a dispute, and it's the thing people never keep.
  • Identified my actual rights. Not legal advice — but a clear-eyed read of what protections and appeal steps existed for a situation like mine, so I knew what to ask for instead of guessing.
  • Drafted the appeal, then the complaint. When the internal appeal stalled, we drafted a formal complaint to my state's department of insurance. Structured, factual, dated, attached to the timeline. The kind of document that's hard to ignore.

None of these are things I couldn't have done alone. They're things I wouldn't have done alone, because each one is a small wall of friction, and stacked together they're the reason people quit.

AI didn't replace my effort. It removed the part of the effort that has nothing to do with being right — the formatting, the translating, the organizing — so the only thing left was showing up.

The regulator step most people skip

Here's what I didn't fully appreciate until I was in it: you are not stuck with your insurer's decision.

Insurers have an internal appeals process, and you should use it. But above that sits a state regulator — a department of insurance — whose job is to take complaints from people exactly like me. Almost nobody uses it, partly because almost nobody knows it exists, and partly because "file a formal complaint with a state agency" sounds like it requires a lawyer.

It doesn't. It requires a clear story, a clean timeline, and the patience to write it down correctly — and that last part is precisely where AI changes the math. Drafting a structured regulatory complaint used to be an afternoon of dread. With Claude it was a focused hour of me directing and editing while the machine handled the scaffolding.

The filing is in. I don't know yet how it resolves. But it's filed — and it wouldn't be, without tooling that made persistence cheap enough to be worth it.

What this means if you're not technical

I'm not a lawyer and I'm not a coder. I'm an operator who has learned that most "I can't deal with this" problems are really "the homework gap is too wide" problems. Insurance is the perfect example. The fight isn't hard. The fight is tedious. And tedium is exactly what you should be handing to AI.

The mental model: AI does not make you right. If your claim genuinely isn't covered, no amount of clever drafting changes that. But when you are right and you're being worn down by process, AI levels the asymmetry. One person with a legitimate grievance can now match the institutional patience of an organization built to outlast you.

For the price of an hour and some attention, I turned a bill I almost paid into an open case with a regulator. Win or lose, that's a different relationship with the system than the one most of us settle for. The next time a denial letter lands, try treating it as a project instead of a verdict.

If you've used AI to push back on something bureaucratic — insurance, a billing error, a benefits appeal — drop it in the comments. And if you want to talk through a fight like this, reach out through the contact page. These are some of my favorite conversations.

Comments

Loading comments…

Leave a comment

0/2000

— Read next
AI & WorkJune 12, 2026 · 5 min read

AI SEO Is the New Visibility Game. Here's How I Picked a Tool For It.

People are asking AI instead of Googling. That quietly rewrites the rules of being found. Here's how I evaluated Profound versus Athena for work — and why we went with Athena.

AI & WorkMay 20, 2026 · 5 min read

I Tried the Big AI Note-Takers. I Keep Coming Back to Granola.

I ran the major AI meeting-notes tools through real work — including a head-to-head with the Gemini note-taker built into Google Meet. One quietly won. Here is why Granola earned the spot.

AI & WorkMay 12, 2026 · 5 min read

Opus 4.8 Is My Daily Driver. Then Fable Cracked a Problem It Couldn't.

Claude Opus 4.8 is a real step up and my everyday workhorse. But the first time I handed a sprawling, multi-step problem to Fable — a long-horizon model — I saw where this is actually going.