Writing and content decision
Rephrase
A capable technical user can build a useful one-user replacement in about a week using existing LLM APIs and common web stacks; the main losses are the vendor's curated presets, extension polish, and any proprietary tuning or SLAs.
Visit website↗Built by Benjamin | Product Builder, who ships 4 products in this index
Not priced
No pricing page we fetched carried a figure, so there is nothing to compare against. The build side is still real.
$100one-off34 h to build
$20/mo3 h/mo upkeep
No published price to break even against.
No open-source build does this yet
Nothing published replaces this one, so a replacement starts from an empty file. Here is what it would have to cover.
What a replacement has to do
- Accept pasted AI-generated French text, select a style/preset or user-specific "Ma plume", call an LLM to rewrite the text into publication-safe French, and return the humanized result for copy/paste.
What it still won’t have
- The vendor-curated library of 28 native French presets
- Their "Ma plume" trained/personalized behavior and any proprietary tuning
- Chrome extension and any integrated UX polish
- Vendor assurances about GDPR/UE hosting and AES-256-GCM claims unless self-hosted accordingly
What remains hard
- Product polish and ongoing maintenance
First-year cost
No published price
Rephrase does not publish a price we could read, so there is nothing to compare against. What building costs is below.
Money you would actually spend
Time you would spend
—
What you would spend
What we assumed
The verdict above measures whether you could build it. This one is only about money.
Runnable build prompt
Build a single-tenant web service (React frontend, Node.js/Express backend, Postgres, deploy to an EU VPS or VPS-like host) that humanizes AI-generated French text. In scope: paste-input UI with word count, preset selector (admin-managed 20–30 presets), user account and storage of 3–5 example texts for a personalized "Ma plume", server-side LLM integration (OpenAI/Anthropic) with prompt templates to neutralize generator-specific tics, result post-processing to strip invisible characters and enforce UTF-8 typographic safety, simple Chrome extension endpoint to send selected page text, AES-256 encryption for stored user examples, and basic GDPR-compatible privacy controls. Out of scope: training proprietary models, building a marketplace, multi-tenant billing, and advanced analytics. Include error handling, rate-limiting, automated tests for prompt-output correctness and input validation, and CI deploy scripts.
How we checked
How the score was reached
- Build verdict base78
- Evidence score78
The base comes from the verdict. Everything under it is a check that either happened or did not, and each one is a fact frozen in this record rather than a judgement made at render time - so the same evidence always produces the same number.
How scoring works →Integrity checks
What held up, and what did not.
