Product
Medical Bill Decoder
Paste the line items from a hospital, clinic, lab, imaging center, ambulance, or pharmacy bill exactly as printed, one line per row in the format 'CODE | description as printed | charge amount'. Get back a plain English translation of every CPT and HCPCS code (with a confidence chip that says 'well known' or 'verify with provider' when the tool is not certain), a typed list of billing error patterns framed as call warrants (never as proof of overcharge), and a bill review call kit: the questions to ask the billing department in order, a first person phone opener the patient can read aloud, and the state specific escalation agency (a Consumer Reports poll found 87 percent of Americans do not know theirs). Every flag is 'this warrants a call', never 'this is an overcharge'. Not legal, financial, or medical advice.
About this tool
What does medical bill decoder do?
Paste the line items from a hospital, clinic, lab, imaging center, ambulance, or pharmacy bill exactly as printed, one line per row in the format 'CODE | description as printed | charge amount'. Get back a plain English translation of every CPT and HCPCS code (with a confidence chip that says 'well known' or 'verify with provider' when the tool is not certain), a typed list of billing error patterns framed as call warrants (never as proof of overcharge), and a bill review call kit: the questions to ask the billing department in order, a first person phone opener the patient can read aloud, and the state specific escalation agency (a Consumer Reports poll found 87 percent of Americans do not know theirs). Every flag is 'this warrants a call', never 'this is an overcharge'. Not legal, financial, or medical advice.
How does Medical Bill Decoder work?
Medical Bill Decoder reads a US medical bill you paste. Paste each line item as it appears on the statement (date of service, code, description, and amount), name your US state, and pick a tier. One structured AI pass decodes every CPT and HCPCS code to plain English with an honest confidence marker (high, medium, or verify_with_provider), fires a typed list of billing-error flags that warrant a call (duplicate charges, upcoding, unbundling, pharmacy markup, facility fee, balance billing, missing itemization, quoted-versus-billed discrepancy, out-of-network higher than expected), and builds a bill-review call kit you take into your dispute call: 5 to 10 ordered questions for billing, a first-person phone opener you can read aloud, and the exact state agency that handles medical-bill and health-insurance complaints in your state. You get the decoded bill plus the call script back in seconds.
How is this different from asking ChatGPT to explain a medical bill?
Freeform ChatGPT happily invents a plain-English description for a CPT code it does not actually know, tells you confidently that you were overcharged by a specific amount it cannot verify, and rarely names the exact state agency you would escalate to. Medical Bill Decoder runs one structured pass with four hard rules: it prints a confidence tier on every decoded line and returns verify_with_provider instead of guessing when the code is ambiguous, it never claims a specific overcharge figure because it does not have your provider contract or your insurer's negotiated rate, it frames every error flag as warrants-a-call rather than proof of overcharge, and it hard-codes the correct state escalation authority for the US state you named so you are not chasing a Google result. You also get an ordered question list and a first-person opener line, not just a paragraph of explanation.
Do you tell me if I was overcharged, or claim a specific amount?
No, and that is a deliberate line the tool will not cross. The tool does not have your provider's negotiated contract with your insurer, your plan's allowed amount, your remaining deductible, or the local Medicare fee schedule for your zip code, so it cannot honestly compute a specific overcharge figure. What it does instead is flag typed patterns that historically warrant a phone call: a line billed twice on the same date of service (duplicate charge), a higher-complexity code where the visit sounds lower-complexity (upcoding), separately billed components of a bundled procedure (unbundling), a pharmacy markup that reads unusually high against the drug and dose, a facility fee tacked onto a routine visit, a balance-billed amount above what your explanation of benefits shows as your responsibility, missing itemization on a large summary line, and any billed amount that does not match a written quote you were given. Every flag is phrased as a question you raise with billing, never a claim you can defend without the underlying rate data.
What are CPT and HCPCS codes, and how does the tool know them?
CPT (Current Procedural Terminology) codes are the 5-digit codes maintained by the American Medical Association that US providers use to bill for every procedure, office visit, lab, imaging study, and surgery. HCPCS Level II codes are the alphanumeric codes maintained by CMS for supplies, drugs, ambulance runs, and equipment that CPT does not cover. The tool decodes the codes it recognizes into plain English (for example, 99213 is a mid-level established-patient office visit, 80053 is a comprehensive metabolic panel, J1885 is a ketorolac injection). When a code is ambiguous, recently updated, category-restricted, or the tool is not confident in the plain-English description, the line comes back with confidence tier verify_with_provider instead of a made-up description. That way you always know which lines are safe to read out loud on your billing call and which lines need the provider to describe them in plain English before you accept the charge.
Will it work with a bill from a hospital, clinic, ER, ambulance, imaging center, or pharmacy?
Yes on all of them, as long as the bill is a US itemized statement with line items you can paste as text. Hospital and ER bills usually mix CPT codes for services with HCPCS Level II codes for drugs and supplies, and both are decoded together. Ambulance bills use HCPCS A-codes for the transport type and mileage, and those decode as well. Imaging center bills are usually a small CPT set (76700-series ultrasound, 71046 chest x-ray, 74177 CT of the abdomen and pelvis with contrast, and so on), and imaging is where quoted-versus-billed and facility-fee flags fire most often. Pharmacy bills use NDC codes for the drug plus a dispensing fee, and pharmacy markup is a first-class flag category. A summary bill with no line items still runs, but it fires the missing-itemization flag near the top of the call kit so you can request an itemized statement from the provider before you argue anything else.
What is the 'verify with provider' confidence tier and why does it matter?
Every decoded line comes back with one of three confidence tiers: high (the code and description are well established and unambiguous), medium (the code is known but the description depends on modifiers or context you did not paste), and verify_with_provider (the code is ambiguous, recently updated, category-restricted, or the tool is not confident enough to describe it in plain English). The verify_with_provider tier is the honesty marker: it means the tool refuses to invent a description just to fill the row. That matters because up to 80 percent of US medical bills have been reported to contain errors (Umbra Health Advocacy, Gitnux), and you cannot dispute an error effectively if the tool made up the description of the line you are disputing. When you see verify_with_provider on a line, ask billing to describe it in plain English before you accept the charge or agree to a payment plan built on top of it.
What is the state escalation authority named at the end, and is this legal or medical advice?
When you enter your US state, the call kit names the exact agency that handles medical-billing and health-insurance complaints for you: usually the state Department of Insurance or Department of Financial Services, sometimes routed through the Attorney General's health-care bureau, and for facility-billing complaints often the state Department of Health. This is baked in because 74 percent of insured Americans do not know their state appeal path exists and 87 percent do not know which agency handles health-insurance complaints (Consumer Reports). Having the agency named on the sheet turns escalation from a Google chase into a phone number you can call. This tool is not legal advice and it is not medical advice. Treat the output as a preparation and escalation kit only. It does not replace a patient advocate, a healthcare attorney, your physician, your pharmacist, or your insurer's member services line.
Which tier should I pick, Quick or Deep, and how much does one decoded bill cost?
There is no subscription. Quick is 15 produde tokens and runs on Gemini 2.5 Flash Lite: it decodes every line, prints confidence tiers, fires the full error-flag taxonomy, and returns the call kit with the state escalation authority. Pick Quick for a short bill (an urgent care visit, a single imaging study, a pharmacy statement) or a first pass on a bill you want to understand before your call. Deep is 30 produde tokens and runs on Gemini 2.5 Flash: it does the same work with more careful code disambiguation on ambiguous CPT and HCPCS entries, sharper judgment on which flags actually warrant a call versus which are artifacts of how your plan pays, and a more specific call-script opener tailored to your exact bill. Pick Deep for a hospital or ER itemized statement, any bill with roughly 20 line items or more, or when the amount at stake is large enough that a wasted 20-minute billing call is expensive. Tokens are a prepaid platform credit you buy once and spend across any produde tool, and if a run fails the full token cost is refunded to your wallet automatically.
What does it cost?
From 15 tokens. Pick a tier on the form above. Produde tokens are a prepaid platform credit; buy them once and spend on any tool.
Do I pay if a run doesn't complete?
No. You only pay for results. If a run can't finish, its full token cost stays in your wallet automatically.
Is my data stored?
Your input and the run output are saved to your account so you can browse history. They are not sold, shared with advertisers, or used to train shared models. See privacy for details.