Pull requests, but for the things you write.
Branchpost is a small SaaS that connects to a GitHub repo, learns your publication context, and drafts blog posts that open as pull requests you review. Same workflow you already use for code — just pointed at the content directory.
Why we built it
Every founder we know has the same blog problem: a repo full of product work that would make great posts, a long list of topics nobody is touching, and a stack of generic AI writing tools that need so much prompting they are not worth opening.
Meanwhile, the same teams have already solved the harder problem — reviewing text — for code. Pull requests are the most underrated content workflow on the internet. They diff well, they tolerate iteration, they nudge you toward quality before publishing, and they keep human judgment in the loop.
Branchpost is the boring observation that you can point that machinery at writing. We read what your repo says about your product, what your existing posts say about your voice, and what you tell us about your audience. Then we draft a post and open it as a PR. You merge or you do not — same as code.
Four things we will not move on
AI writing needs a review surface, not a publish button.
Code already solved this — pull requests. Most AI writing tools dump text into a chat window and hope you copy-paste it somewhere. Branchpost opens every draft as a PR you can read, comment on, edit, and merge.
Context beats prompt engineering.
The reason AI drafts feel generic is not the model — it is the input. Your audience, positioning, tone, and existing posts are the real prompt. We save that context once and reuse it on every draft.
Your repo is the source of truth.
No CMS, no proprietary lock-in. Branchpost reads the structure you already have (MDX in a content directory) and writes back into it. Cancel anytime — your posts live in git, where they always did.
Predictable spend over per-token surprises.
Credits, not tokens. Monthly resets, optional top-ups that never expire. You should know what a draft costs before you generate it.
I have been around enough “just point AI at it” products to know the gap is not the model. The gap is everything around the model: the context you feed it, the surface where you review the output, and the trust you build by never publishing something behind your back.
Branchpost is opinionated about all three. If those opinions match how you already work, we would love to draft your next post.
— Beka, founder