Legal
Disclaimers
Last updated: August 30, 2026
How Benchmark's reference library was assembled
Every document in Benchmark's reference library was collected from publicly accessible sources on the open web — manufacturer collision repair sites, manufacturer technical resource pages, and the OEM Collision Repair Roundtable at oem1stop.com. Each record carries the URL it came from and the revision date printed on the document itself.
Nothing in the library came from a paid subscription service. No content from ALLDATA, Mitchell, or any manufacturer's licensed service portal has been uploaded to or incorporated into it. No account was created and no access control was circumvented in assembling it.
The library is historical and periodical — documents of varying ages, superseded revisions kept for the record, and manufacturer bulletins published over many years. It is a reference collection, not a substitute for current licensed service information.
Where a manufacturer keeps a procedure behind a subscription, Benchmark can say where it lives. It will not supply it. Information a facility has not paid for is not Benchmark's to hand over, and verifying current service information for a specific vehicle remains the responsibility of the facility performing the repair.
How Benchmark uses AI, and what it does not do
Benchmark Diagnostics does not develop AI. Benchmark Diagnostics is a diagnostic and calibration company. It does not build, train, or fine-tune AI models, it does not analyse your customers or your business, and it will not. The AI models it uses are commercially licensed, and this platform is developed and maintained independently by Sidney L. Dunn Jr.
Your information is never used for development. The estimates, photos and repair plans a facility submits — and the customer and claim information printed on them — are read to prepare the review that was asked for, and are used for nothing else. They are not training data, here or anywhere.
The reference library is read, not absorbed. When a review is prepared, the manufacturer documents relevant to that vehicle's make are supplied to the AI for that single request and are not retained by it afterward. This is the same act as a technician opening a service document — performed faster, on more documents at once.
A person finalizes every review. The AI prepares; a Benchmark technician reviews, corrects and publishes. Nothing reaches a repair facility without a person having looked at it.
How the tools improve. When a Benchmark technician corrects the AI's output, that decision is recorded. Those records describe judgments about vehicles and repairs. They are not a collection of your customers, your claims or your business. They are provided to Sidney L. Dunn Jr., who develops these tools independently of Benchmark Diagnostics, and are used to improve them.
What we do with the documents you upload
Benchmark is not an auditor and does not police what appears on the documents you send. We minimize what we keep because it is the right way to handle someone else's customer, and because facilities work under agreements with carriers about how that information is handled. Keeping less makes that easier for you.
The documents themselves are deleted. What we keep is the work prepared from them.
Estimate review — deleted after 24 hours. Estimates, repair plans and photos uploaded by repair facilities are deleted 24 hours after submission, along with their file names.
At the same time, the vehicle identification number is shortened to its first eleven characters. Those characters describe the year, make, model, plant and body — everything a repair plan depends on. The remaining six are the serial that identifies one particular vehicle, and we discard them.
Operation descriptions read from the estimate are retained as part of the review. They are filtered against customer and claim content, but that filter is a safeguard rather than a guarantee. Information printed inside an operation description on the source estimate can persist with that description.
The review itself is retained in full, so nothing about your repair plan is lost. A facility can re-upload a source document at any time.
One exception worth knowing about: the notes field a facility fills in when submitting a review is kept as written. It is not filtered and it is not purged. Anything typed there persists, so it is the one place worth leaving customer and claim details out of.
The comparison tool is automated and unreviewed
The free estimate comparison tool is fully automated. No Benchmark technician reads the two estimates, checks the matching, or approves the worksheet before it reaches you. It is not an estimate review, and it is not Benchmark's professional opinion about a repair.
What it does. It reads the two documents you upload, matches operations between them where it can, and reports what appears to be added, removed, changed, or unchanged. Matching is a judgment about wording, and wording differs between estimating systems. Operations can be matched wrongly, missed entirely, or paired when they are not the same work.
Money figures are read from the documents. Totals, rates and line amounts come from the documents as printed and are reproduced without verification. A figure the tool could not read is reported as unread rather than guessed. Check every dollar figure against the source estimates before relying on it.
It does not decide what a vehicle requires. The tool compares two documents to each other. It does not consult manufacturer service information and it makes no statement about whether either estimate is correct, complete, or compliant with any procedure. Determining what a repair requires is the work of the facility performing it — and of a Benchmark estimate review, which is prepared against manufacturer documents and finalized by a technician.
Use the worksheet as a starting point for your own review of the two documents. Benchmark is not responsible for decisions made on the output of the free tool without that review.
The estimate PDFs you upload are never stored. They are read to produce the comparison and are not kept.
What a run keeps is the record of the comparison itself — the matched operations, part numbers and hours, each document's printed total, and the difference where operations matched. The vehicle identifier is shortened to its first eleven characters, which describe the year, make, model, plant and body and identify no particular vehicle.
The record never contains either party's labour rate or the dollar amount on any individual line. Those appear only on the priced worksheet handed back to you, which we never store.

