Less like AI.
More like you.
Two APIs for teams that publish under a real name. Rewrite moves AI-generated drafts toward your writer's habits and phrases, so they sound more human and more like you, and checks every number, name and link against the source. Detect shows, sentence by sentence, why text reads as machine-made.
- 10,000 free credits every month
- No card to start
- REST, webhooks, TypeScript & Python
POST /v1/rewrite
fidelity: strictIn today's fast-paced world, getting paid on time matters more than ever. That's why we're excited to introduce Smart Reminders, a game-changing way to streamline your invoicing. Invoices unpaid after 14 days are now escalated automatically. In our beta, teams recovered $32,960 in late payments. Ultimately, it's not just about reminders, it's about building lasting client relationships. Learn more at intactvoice.com/reminders.
Late invoices were eating our customers' weeks. So reminders escalate on their own now: anything unpaid after 14 days gets a firmer nudge. Beta teams got $32,960 in late payments back. Details at intactvoice.com/reminders
—
AI-likelihood
Signals
Fidelity
- 14 days
- $32,960
- link
Built for teams that publish under a real name.
The platform
Everything between a draft and publish.
Two endpoints do the heavy lifting. Voices, fidelity checks, presets and async jobs make them safe to run on every post you ship.
Fidelity checks
A rewrite that changes a fact is a bug.
Every number, unit, name and link in your source is checked against the output. If one drifts, the request fails instead of shipping a different claim.
Source · support policy
3 protected factsEscalate any ticket older than 14 days to a team lead. Credits above $500 need finance approval. Full policy: intactvoice.com/support-policy.
A typical humanizer
200 OKTickets that have been sitting for more than 14 weeks should go to a team lead. Credits over $500 need a sign-off from finance. See intactvoice.com/support-policy.
IntactVoice · fidelity: strict
Anything open longer than 14 days goes to a team lead. Credits over $500 need finance to approve them first. The full policy is at intactvoice.com/support-policy.
number_unit
14 days
currency
$500
url
link
200 OK · fidelity.status: "pass"
“14 days” → “14 weeks” is a real output from a humanizer we tested. Source: our lab test, Oct 2026. Example text rewritten for illustration.
Detect API
Detection that shows its work.
A single score tells an editor nothing. Detect returns the sentence, the pattern and the evidence, and treats all of it as a signal, not a verdict.
POST /v1/detect
likely_ai · 0.91In today's fast-paced business landscape, onboarding has never been more important.
In this post, we'll explore three strategies to help your team get new hires productive faster.
First, set up accounts before day one so nobody spends a morning waiting on IT.
Second, pair every new hire with a buddy for their first two weeks.
Third, write down the ten questions everyone asks, and answer them once.
By implementing these strategies, you can create an onboarding experience that drives engagement and retention.
In conclusion, great onboarding isn't just a process, it's an investment in your people.
Per-sentence AI-likelihood
4 of 7
sentences flagged
3
read like a person
medium
confidence · 96 words
Voice profiles
Not “friendly but professional”. Measured.
Give us two to twenty things a person actually wrote. We measure how they write (sentence spread, punctuation, how they open and close) and reuse it on every call.
X post · sample 1 of 6
Shipped the invoice export. 4 days late. Worth it: CSVs now open in Excel without mangling the dates.
Blog · sample 2 of 6
We tried weekly planning for a month. It didn't stick. What replaced it: one doc, three columns, updated on Fridays.
Team email · sample 3 of 6
Quick status: migration is 80% done. Two tables left. I'll post again Thursday.
Founder voice
voice_8f2k1x · 6 samples · 4,210 words
Sentence length
sd 9.4 words · wide spread
Punctuation / 100 words
- Period8.1
- Colon1.4
- Comma3.2
- Question0.3
- Em dash0.0
Opens with
a verb or a fact
Ends on
a status update
Reveals with
a colon
Detect baseline
0.19
Research, cited
Word lists are the weak signal. Structure is the strong one.
The product is built on published work, and the claims on this page come with their sources. Here is what the field has measured so far.
0.0
macro-F1
Structure alone separates AI from human B2B blog posts. Still 96.1 after the AI rewords its own posts.
SlopShape v3, arXiv 2609.15369, Sept 2026
0.0%
caught
Readers who use AI heavily spotted AI articles with 4.0% false positives. Everyone else did no better than a coin flip.
Russell, Karpinska & Iyyer, ACL 2025
0%
false flags
Share of essays by non-native English writers that perplexity-based detectors labelled AI. Why we calibrate to the writer.
Liang et al., 2023
70.3% → 0.0%
detection
One paraphrase pass and DetectGPT's hit rate at 1% false positives collapses. Word-level detectors are easy to fool.
Krishna et al., NeurIPS 2023
5.3×
GPT-4o's rate of trailing “-ing” analysis clauses vs human writers
Reinhart et al., PNAS 2025
28×
“delves” in 2024 PubMed abstracts, vs its expected rate
Kobak et al., Sci. Adv. 2025
~40%
fewer views for LinkedIn posts flagged “seems like AI slop”, per LinkedIn
LinkedIn, via TechCrunch, Jul 2026
2 Aug 2026
EU AI Act Article 50 transparency duties apply
Regulation (EU) 2024/1689
The takeaway we build on: fixing words doesn't fix AI writing. Fixing structure, rhythm and voice does, and so does checking that facts survive the edit.
Read the data sheetsFor developers
Your first call in three steps.
One REST endpoint per job, JSON in and out, and responses that explain themselves. Plain HTTPS from any language.
- 1
Create a key
Sign up, open the dashboard, copy a test key. 10,000 free credits a month, no card.
- 2
Send a draft
POST the text with a voice_id and a platform. A job id comes back at once; poll for the result or receive a webhook when it finishes.
- 3
Read the receipts
Every finished job carries the fidelity checks it ran and before/after Detect scores.
curl -X POST https://api.intactvoice.com/v1/rewrite \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{
"text": "Many businesses struggle to create useful blog posts because the writer receives only a topic. Escalate any case older than 14 days.",
"voice_id": "voice_8f2k1x",
"platform": "blog",
"strength": "standard",
"fidelity": "strict"
}'{
"id": "rw_01JD4Q8ZK3",
"object": "rewrite",
"model": "rewrite-2026-10",
"duration_ms": 34210,
"text": "Most blog posts go wrong before anyone writes a word: the writer gets a topic and nothing else. If a case is older than 14 days, escalate it.",
"fidelity": {
"status": "pass",
"checks": [
{
"type": "number_unit",
"source": "14 days",
"output": "14 days",
"ok": true
}
],
"changed_facts": []
},
"signals": {
"source": {
"ai_likelihood": 0.91
},
"output": {
"ai_likelihood": 0.14
}
},
"usage": {
"words_in": 22,
"words_out": 28,
"words_billed": 22
}
}Pricing
Pay per word. Nothing per seat.
One credit rewrites one word. Detection costs a fifth of that. Top-ups never expire, and failed fidelity checks are never billed.
Starter
$49/mo
300k credits · ≈ 300k rewritten words
- Rewrite, Detect & Voices
- 3 voice profiles
- Email support
Growth
$149/mo
1.2M credits · ≈ 1.2M rewritten words
- Everything in Starter
- Higher rate limits
- Usage by key and project
Scale
$499/mo
6M credits · ≈ 6M rewritten words
- Unlimited voices
- Prepaid top-ups
- Priority support
Free: 10,000 credits a month. Paid plans bill monthly. Higher volumes: talk to us. Compare plans →
Publish words you'd sign.
Start with 10,000 free credits. Bring a draft and two samples of the writer, and see the difference on your own content.
- No card to start
- Failed fidelity checks aren't billed
- Cancel any time