I opened my inbox and froze: a perfectly polite email that read like a slide deck written by Bryce. It landed with the clinical hum of an algorithm and the wrongness of someone who rehearsed being casual. You feel that instant suspicion—was this human, or an AI faking humanity?
Meet Sinceerly, a tiny app whose entire promise is to make your AI-written messages read less like an LLM and more like someone who forgets an apostrophe now and then. Its pitch is simple: introduce human-style imperfections—typos, fragmentary phrasing, the occasional clipped sentence—so your emails don’t scream “generated.”
My colleague forwarded three versions of one paragraph—Subtle, Human, CEO—and I couldn’t stop reading how they differed
I fed Sinceerly a deliberately soulless paragraph to see what would happen. The app returns three tones: Subtle smooths some rhythm and adds small contractions; Human leans into casual gaps and a typo or two; CEO strips capitals and trims sentences into a clipped, exec voice. The results are oddly specific: small errors become signals of authenticity, like a magician’s misdirection that steers attention away from the algorithm.
How does Sinceerly work?
It runs a transformer over your text and deliberately inserts human marks—misspellings, uneven sentence length, punctuation quirks—and offers sliders for intensity. You pick Subtle, Human, or CEO, and the model rewrites along those axes. It’s not avoiding AI; it’s weaponizing another AI to simulate human fallibility.
At a demo the developer laughed and called it “tongue-in-cheek,” which felt both honest and strategic
Dan Horwitz, a genial MBA student at Harvard, told me the project started as satire. That deadpan homepage—overly serious about making mistakes—works as a mirror. People ask whether the joke lands or backfires: the site intentionally flirts with the idea that polishing away every quirk makes you less trustworthy, not more.
Is Sinceerly satire or a real product people will use?
The short answer: both. Horwitz says it’s meant to make you smile and think; teams will still use it because reputation matters. You can imagine a marketing manager running text through ChatGPT, then feeding the results into Sinceerly before sending to clients or pasting into a LinkedIn message.
I watched a recruiter toss out a perfect pitch because it didn’t sound human, and that changed my thinking
We’re at the strange point where perfect prose signals machine authorship. Tools like Grammarly, ChatGPT, and Sinceerly are now part of a feedback loop: an LLM writes, an app humanizes, another LLM reads. It’s a hall of mirrors where each reflection is slightly warped.
This matters because trust in writing isn’t only about grammar; it’s about perceived intent. If every message reads like a rehearsed commercial, people start to prefer the imperfect email from someone they trust. That preference is fertile ground for products that add the right kind of mess.
You can scoff and call it parody—or you can imagine the middle manager who runs a quarterly outreach through this stack and avoids sounding like a corporate bot. Either way, the technology forces a question about authenticity: are we solving a cosmetic problem, or are we reshaping honesty in digital communication?
So, will you let an app add the human scars to your prose to save a meeting, a deal, or your reputation?