Every other AI resume tool will invent your accomplishments

It won't write anythingyou can't prove.

Paste a job description into any AI resume builder and it will hand you a number you never hit, a tool you have never opened, a title you never held. TruthResume works the other way around: you supply the evidence first, and every line it writes has to trace back to something you actually did. When a claim cannot be supported, the engine blocks it instead of shipping it.

Private preview. The demo is one verbatim engine run — nothing on it is a mockup.

No logo wall. Here is the engine instead.

Watch it refuse.

Two bullets, one engine, one set of rules. The only thing that differs is whether the evidence is there.

BLOCKEDexit 1

“Led a team of 40 engineers to revolutionize the deployment pipeline, leveraging AWS Solutions Architect certification.”

  • 01no-invented-numbersThe figure 40 appears in no evidence file.
  • 02vocabulary-must-come-from-evidenceSupport ratio 0.18, against a 0.60 threshold.
  • 03no-unlisted-credentialsA certification claim with nothing behind it.
PASSexit 0

“Cut the nightly reporting job runtime from 90 minutes to 22 minutes by rewriting aggregation queries and adding partition pruning.”

Every figure and every claim traces to a line in the evidence. Nothing to flag, so it ships.

The non-zero exit code is the point: you can put this in a git hook or CI and fail a build on a fabricated claim.

How it holds

Three things no resume builder does.

Your evidence, in a file you own

A truth pack is a directory holding what you have done, the rules you want enforced, and the formats you like. The engine knows nothing about you beyond it. Import it from a resume you already have — there is no config file to write.

Lock a line by highlighting it

Open your resume in Word and paint the parts that must never be reworded — your degree, your dates, your publications. Highlighted text is reproduced exactly, forever, including link styling. Highlight half a sentence and only that phrase locks.

Rules that only ever tighten

Your standards are declarative rules, never code, so a pack can be shared without executing anything on anyone's machine. A pack can reject a line the engine allowed. It can never rescue one the engine rejected.

How it works

Three steps, and a gate at the end.

  1. 01

    Import

    Point it at the resume you already have. Your highlights become locks.

  2. 02

    Tailor

    Paste a job posting. It drafts only from evidence you supplied.

  3. 03

    Verify

    Every line is gated before it can reach a document. Unsupported claims are blocked with a reason.

Honest status

A product about not overstating things should not overstate itself.

This is a private preview of a working engine, not a finished service. Here is exactly where each piece stands.

CapabilityStateDetail
Truthfulness gateWorkingBlocks unsupported claims with itemised findings.
Truth packsWorkingEvidence, rules and templates are user-owned and portable.
Import from WordWorkingYour highlights are read as lock declarations.
Import from PDFIn progressA PDF carries no highlights, so locks become defaults you review.
ATS-aware outputWorkingKeyword scoring, gated so it can never invent a keyword.
Hosted web appWorkingImport, lock, tailor and build, all in the browser. Private preview; public deployment is next.
Pricing

Pay once. Nothing renews.

Every other resume tool bills you monthly until you remember to cancel. This one sells a batch of builds and then leaves you alone - the same principle as the engine itself, applied to your card.

Free NO CARD

$0

Three tailored resumes, no card required. Enough to judge whether it is any good.

Credits PAY ONCE

$25for 20 builds

About $1.25 a resume. They do not expire, and no subscription starts.

Payment happens on Stripe's hosted page; your card never touches this service, and the billing record here holds no personal information at all.

Questions

The things people actually ask.

Does it use AI?
Optional, and off by default. The gate itself is deterministic rules, not a model, so the same input always produces the same verdict.
What stops it inventing things?
Every line is checked against your evidence before it can reach a document. Unsupported claims are blocked, and you are told which rule fired and why.
Can I keep my formatting?
Yes. Anything you highlight is reproduced character for character, including link styling. No renderer or optimizer may touch it.
What if it blocks something true?
Add the supporting evidence to your pack and it passes. The block is about what you gave it, not a judgement about you.

See it hold the line.

The hosted app is in private preview. Until it opens, the demo shows one complete, verbatim engine run - the import, the locks, an AI proposal accepted, another refused with its arithmetic, and the delivered document.

See a real buildReal output, no mockups.