Voice profiles now travel with the team.
A voice profile trained by an editor in Berlin now applies automatically to every draft opened by a teammate in São Paulo. Version history, comments, and tone rules stay locked to the profile, not the seat.
Issue No. 04 — A field guide to AI that sounds like you
The AI writing partner that thinks in your voice — not its own.
Nulis is the only AI writer trained on your existing content library. Every draft reads like a senior member of your team wrote it, not a chatbot with a personality. Marketers ship 3.4× more long-form pieces per month without losing the editorial voice their audience trusts.
No credit card. Built for content teams shipping 20+ articles a month.
A note from the founders
We left The Verge and Wired to fix the one thing every AI writer kept getting wrong.
For ten years between us, we edited stories for readers who notice when a sentence is off by a syllable. We know what a publication's voice feels like from the inside — the rhythm, the hesitations, the verbs a writer chooses, the jokes that land and the ones that don't. When the first wave of AI writers shipped in 2022, we tried them on our own drafts. The output was clean. It was also completely anonymous.
Nulis started in a Brooklyn apartment in 2021, two years before any of those tools had a name. Our bet was simple: a model that reads twelve of your published pieces will learn your voice faster than a generic system that has read the entire internet. Eight thousand two hundred content teams later, including editorial groups at Notion, Linear, and Vercel, that bet has held up. We still ship a meaningful release every 7.8 days. We still hire editors before engineers.
— Hana Olu & Marc Reyes, co-founders, Nulis, Inc.
The Voice Engine
Teach it twelve pieces of your writing. It will sound like your team on the thirteenth.
Most AI writers need fifty or more samples before they stop sounding like a chatbot. Nulis locks onto a publication's tone from a single shelf of archived work — and keeps that voice intact across every format you publish in.
-
01
Drop in your archive
Paste twelve published URLs or upload a folder of articles. Nulis reads them the way a new senior hire would — for cadence, vocabulary, sentence length, and the small editorial tics that make a publication feel like itself.
-
02
Train the model in an afternoon
A voice profile is ready in under four hours. Compare a fresh draft against your archive side-by-side, tweak the rules of thumb the model inferred, and lock the profile to your whole team. No data leaves your workspace.
-
03
Ship drafts that read like your team wrote them
Briefs, long-form features, newsletter sends, product pages — every draft arrives pre-styled in your house voice. Average editing time drops 62% in the first thirty days, measured across 3,400 onboarding sessions.
What the receipts say
Four numbers we are willing to defend in front of any editor.
- #1 Ranked on long-form coherence by Stanford's CRFM, March 2024 evaluation.
- 3.4× More published long-form pieces per content team, measured 90 days after onboarding.
- 62% Reduction in editing time within the first 30 days, audited across 3,400 sessions.
- 99.4% Of first-draft output passes Copyscape in our independent originality audit.
SOC 2 Type II certified since November 2022. GDPR compliant since launch. Backed by $14.2M Series A led by Andreessen Horowitz.
Where the work goes
Draft in Nulis. Publish where you already write.
Nulis is the only AI writer with native one-click publishing to WordPress, Webflow, and Ghost — no copy-paste step, no reformatting the headings, no broken image embeds. Drafts land in your CMS exactly the way your editor expects to find them.
- WordPress — push a finished article straight to a draft post, complete with your featured image and slug conventions.
- Webflow — populate CMS items from a Nulis brief, including SEO fields and Open Graph metadata.
- Ghost — publish to a draft or schedule a newsletter send without leaving the editor.
Start the 14-day free trial
No credit card. Cancel any time before day 14.
The 2024 shipping log
Forty-seven releases in one year. Here is what they actually changed.
We average a meaningful product release every 7.8 days. Not patches, not renames — releases that move the editorial floor. A short selection from the 2024 changelog, edited for the readers who care about the work.
Brief mode for SEO teams.
Briefs can now be generated from a target keyword cluster, a competitor set, and your existing pillar pages. The output is a working outline calibrated to your archive — not a generic content brief that ignores what you have already shipped.
One-click publish, for real this time.
Native publishing to WordPress, Webflow, and Ghost moved out of beta. Drafts land with headings, slugs, featured images, and SEO fields already mapped to your conventions. Editors stopped copy-pasting in Q4.