Paste a claim and it tells you why a payer would reject it, then repairs what it can derive. Paste a remittance and it tells you what each denial actually means, whether it is worth appealing, and what to do next. Patient identifiers are removed before anything reads the file, so nothing you open here is PHI.
Recovering a denied claim takes a call, an appeal and a month. Catching it before submission takes one click. Built for the people who read raw X12 at eight in the morning because a batch failed overnight.
One hundred rules, each with a code and a plain-language reason.
Envelope counts, control-number pairing, required loops, NPI check digits, ICD-10 qualifiers, diagnosis pointers that reference codes which do not exist, and the claim total against the sum of its service lines. Every finding traces back to the exact element it came from.
Thirteen fixes derived from the file. Five that ask you for the one fact only you have.
Segment recounts, control-number alignment, retired qualifier upgrades and element realignment happen on one click. A five-digit ZIP or a missing diagnosis is never guessed, because a claim that is wrong is worse than a claim that fails.
Identifiers are removed before anything reads the file, and there is no setting that turns it off.
Redaction runs at the tokenizer boundary, ahead of the parser, the validator, the search index and the payer profiler. It is positional rather than content-matching, it keeps no mapping table, and it has no setting, so it cannot be left off by accident.
Paste an 835 and every adjustment is decoded, ranked by dollars, with a next step.
What the denial actually means in plain language, whether it is appealable or a benefit decision you cannot argue with, and what the documentation needs to show. Load your fee schedule and it finds lines paid below contract. Load the CMS edit table and bundling denials show whether a modifier is permitted at all — which is as often a reason not to append one as a reason to.
Most tools redact on the way out. That is too late, because the parser has already read the data. This one redacts between the tokenizer and everything downstream, and the step cannot be skipped or switched off.
A remittance tells you a claim was denied. It does not tell you what to do about it. Paste one in and every adjustment is decoded, ranked by dollars at stake.
Each adjustment reason explained in plain language, grouped by whether it is an authorization problem, a coverage problem, a coding problem or a benefit decision. Remark codes surfaced alongside, because a generic missing-information denial cannot be worked without one.
Marked as worth appealing, worth checking first, or not appealable at all. A medical necessity denial with good documentation is very different from a benefit exclusion, and working them the same way wastes the time you have.
Load your contracted rates and every line's allowed amount is compared against them. Denied lines are excluded from that total, because a denial is not an underpayment and counting it twice would inflate what you think is recoverable.
Checks the arithmetic at transaction, claim and line level. A remit that does not reconcile will leave a variance when it posts, and finding that before posting is much cheaper than finding it during a month-end close.
Open it here and it runs in this tab. Download it and it runs from your own disk, on your own network, with no dependency on us being online. The code is identical; only the delivery differs. It works on macOS, Windows and Linux because it runs in a browser rather than as a native application.
It lands in ~/Downloads as claim-workbench-1.3.0.html. Safari may unzip or rename it; if so, keep the .html extension.
It opens in your default browser and starts with a sample claim already parsed. macOS does not show a security prompt for HTML files.
Safari is stricter about local files than Chrome. Everything works, but if you plan to import PDFs, Chrome or Edge is the smoother choice.
Move it out of Downloads into Documents or a shared drive, so it is not swept up by storage cleanup.
Edge and Chrome flag downloaded HTML by default. If you see “this file may harm your device,” choose Keep. That warning is generic to the file type, not specific to this file — the checksum on the right lets you verify what you received.
It opens in your default browser with a sample claim already parsed. If Windows opens it in Notepad instead, right-click, choose Open with, and pick your browser.
On locked-down machines, right-click the file, choose Properties, tick Unblock at the bottom, and apply.
Because there is no install, one copy on a network drive works for everyone. No licence server, no per-seat setup.
Save it anywhere. Verify it with sha256sum claim-workbench-1.3.0.html against the checksum on the right.
xdg-open claim-workbench-1.3.0.html, or pass the path to Firefox or Chromium directly.
Disconnect the network and drop a file in. Parsing, validation, repair, redaction and search all run unchanged. Two things degrade: the page falls back to system fonts, and PDF import is unavailable.
The tool tells you why one claim would be rejected. An audit tells you why forty were, and which of them trace back to the same upstream cause.
Run them through the workbench first. Patient names, member identifiers and dates of birth are stripped before the files leave your building. No system access, no integration, nothing to install.
Where the denied dollars went, which payer or provider or procedure is driving each pattern, what is still inside the appeal window, and any line paid below your contracted rate. Written for a practice manager, not an analyst.
A written report ahead of a one-hour findings session, then thirty days of follow-up. We do not touch your systems, contact your payers, or recommend codes — we identify patterns and point at what the documentation needs to show.