Launching today
Split Post
Your voice, every platform, enforced.
3 followers
Your voice, every platform, enforced.
3 followers
Post one idea to LinkedIn, Twitter, and Instagram and you can usually tell which version is actually yours and which two were written by something approximating you. SplitPost generates platform-native variants from one source, then a second AI pass validates every line against your explicit voice rules and rewrites violations before you see anything. Output you copy and post, not output you fix.
I built SplitPost because I was running a business and several side projects at once, and all of them needed a social media presence.
If you want anyone to find your work, you have to post, comment, and engage multiple times a week across multiple platforms. That's the baseline. The sheer volume of content required just to be visible in a saturated market is brutal, and it compounds every week.
So I turned to LLMs to speed some of this up. They helped, until I looked at what they were actually producing.
Every output sounded the same. The em dashes. The "Not A, but B" structures. The rhetorical questions. The performative humility. Different prompts, different models, same result. I was spending more time editing AI output to match my actual voice than I'd have spent writing from scratch.
It got worse when I tried to repurpose. I'd take a long form post and ask for a LinkedIn version, a Twitter version, an Instagram caption. Three outputs, three different writers. The LinkedIn version was corporate. The Twitter version was somebody's idea of how I should sound on Twitter. The Instagram caption used vocabulary I've never used in my life. Same source, same prompt structure, same model. The drift was worse across platforms than within a single output.
I tried every tool. Jasper. Copy.ai. Social Bee. Buffer's AI assistant. They all generate content from a prompt and some tone guidance. None of them enforce constraints. You get output that is approximately right and then you sit there fixing it.
So I built a simple validation tool. A second AI pass that reads the generated output against explicit rules and rewrites anything that breaks one. Not "match this tone." Specific constraints: these phrases are banned, these sentence patterns are banned (each with a BAD example and a GOOD example), contractions required, no emoji, no rhetorical hooks.
That simple tool turned into SplitPost. It takes one piece of content and generates platform-native versions for LinkedIn, Twitter, Instagram, Reddit, and more, each one validated against your full voice profile. The architecture is two AI passes in series: the first generates, the second catches and rewrites violations. The trade off for running two passes instead of one is output you can copy and post versus output you have to read and fix in three places first.
The free tier gives you 3 generations per month with full two-pass validation. No credit card. Enough to test whether the constraints actually produce output that sounds like you.
What I'd ask from this community try it with your own content. Set up a voice profile, run a generation, tell me where the output breaks your rules. Every piece of feedback tightens the enforcement system for everyone.
Sign up with splitpost.io/signup?promo=PRODUCTHUNT to claim your bonus generations.