One pipeline over hundreds of manuscripts, not one paste at a time

Whole files in and out, not a text box. Upload a DOCX or PPTX and get back an editable file in the same format. Layout, figure numbering, reference structure, and in-text citations all survive, so the client does not receive something that has to be rebuilt. Tools that return plain text mean somebody on your side redoes the formatting on every manuscript.

No per-request word cap. Across the category a single request stops around two or three thousand words, on the most expensive tiers too, so a fifty-thousand-word thesis gets split a dozen times or more. We process the file: split semantically, rewritten in parallel, every slice sharing the same source context, so terminology and cross-references stay consistent across the whole document. For bulk work this is the line between automatable and not.

100 credits per 1,000 words, and billed only on words actually rewritten. If the client brings a Turnitin or iThenticate report, importing it confines the work to the flagged passages and freezes the rest - and confines the charge the same way. A report flagging 20% of a manuscript costs 20% of the manuscript, not the whole word count.

Two rewrite strengths, on DOCX and PPTX (PPTX currently runs the balanced strength).

Citation format standardization, at 5 credits per 1,000 words - our cheapest. Client references arrive from everywhere: typed by hand, exported from Zotero, formatted for a previous supervisor. Twelve styles are supported (APA 7, MLA 9, Harvard, IEEE, both Chicago systems, both GB/T 7714 systems, Vancouver, AMA, ACS, OSCOLA), and the body stays frozen.

Condensing to a word limit, at 20 credits per 1,000 words. Hard limits from journals and application portals, worked to a word budget by cutting redundancy and merging sentences rather than dropping content.

Layout-preserving translation, at 10 credits per 1,000 words. PDF, DOCX, PPTX, XLSX, EPUB and more, with figures, formulas, and page layout intact.

The three chain into a pre-submission workflow: humanize first (it changes the word count), then condense to the limit, then standardize citations last - neither earlier step touches the reference list, and once the style is right no rewriting tool should touch it again. Reversed, the work gets redone.

API at https://api.humanpen.net/v1. Upload a file, create a job, poll progress, download the result. API keys are self-served in settings, with read and write scopes.

Or in bulk on the web: if you would rather not integrate, upload directly, run up to three jobs concurrently, and collect the results from History.

Billed on words actually rewritten - not per file, not per seat. Credits are bought once and never expire; failed and cancelled jobs are not charged.

The admin console provides usage detail and settlement export: jobs and consumption over a date range, for internal allocation or client billing. Rates can be agreed separately depending on volume.

We make no effectiveness promise. No guarantee against any detector, and no detector of our own. If your model depends on promising clients they will pass, we are the wrong supplier. What we offer instead is importing the Turnitin or iThenticate report the client’s institution issued, so the verification data comes from the client’s side.

No white-labelling. White-labelling means selling our output as your product, and we make no promise about that output; the two do not sit together. The API integrates into your workflow, but we do not badge-engineer.

The rewrite engine is English-only. Citation format correction is language-agnostic and handles Chinese references, and translation covers 12 languages, but humanizing targets English text only. Chinese support is on the roadmap without a date.

Tell us your type of organization, rough monthly volume, and which feature you mainly need, and we will come back with rates and an integration suggestion: support@humanpen.net