The AI tweet writer for developers who hate writing
Facts checked July 2026
The work is interesting. The audience exists: developer X rewards war stories, benchmarks, and 'here's the gotcha' posts more than any other niche. The blocker is the cold open: turning a merged PR into a post is a genre-translation task, and it's nobody's favorite part of the day. Voxly takes the PR description, the commit message, or the debugging story and turns it into the post you would have written if writing weren't a tax.
It drafts in your voice, learned from your actual timeline, and grounds every claim in the material you paste, so the post keeps its technical spine instead of dissolving into 'I built a thing'.
Sound familiar?
The genre translation tax
You can write a precise PR description in two minutes. Turning it into a post means switching genres entirely, and that context switch costs more than the typing.
Cringe risk
Dev X punishes hype. 'Absolutely blown away by what I just shipped' from an engineer reads as a resignation letter from your credibility.
ChatGPT flattens the details
Generic AI rounds your specific numbers into 'significantly faster' and your gotcha into a platitude. The details were the post.
Great work, zero distribution
The interesting bug you fixed at 2am is exactly what the timeline wants, and it will never leave the incident channel.
Writing feels like overhead
An audience compounds into job offers, users, and leverage. But every minute on prose feels stolen from the build, so it never happens.
Threads take an evening
The deep-dive thread about your architecture decision needs structure, a hook, and pacing. That's an essay, and essays lose to Friday.
How engineers use Voxly
- 1
Voice profile from your timeline
Voxly reads your recent posts and builds an editable profile of how you actually write: dry, lowercase, fragment-heavy, whatever it is. Drafts inherit it. See yours first with the free voice analyzer, no signup.
- 2
Paste the artifact, not an essay plan
The PR description, the commit log, the benchmark table, the link to the release, a screenshot of the flame graph. Voxly drafts the post grounded in that source, so the numbers and the gotcha survive intact.
- 3
Debug the draft, don't rewrite it
Select the weak sentence and say 'skip the framing, jump to the gotcha' or 'keep the exact error message'. Targeted edits, and it remembers your preferences (like: never use exclamation marks) across sessions.
- 4
Publish in one tap
Straight to X through the API when you decide, and only when you decide. No scheduler, no auto-posting, no surprise threads.
Why source grounding matters for technical posts
The failure mode of AI-written dev content is confident vagueness: benchmarks without numbers, bugs without root causes, 'leveraging' where a verb should be. Voxly's generation is reference-grounded: it fetches and reads what you give it (links, tweets, YouTube talks it transcribes, images) and drafts against that material. When you paste a PR that says p95 went from 840ms to 310ms, the draft says that, because making claims traceable to the source is the design goal.
The same applies to threads: give it the design doc or the postmortem and ask for a thread, and each tweet carries a real point from the source instead of padding.
The audience is a career asset; the tax shouldn't be
Engineers with an audience get disproportionate returns: inbound offers, users for side projects, conference invites, negotiating leverage. The reason most engineers don't build one isn't ability (the raw material is sitting in your commit history), it's that the translation step never clears the priority bar. Making that step nearly free is the entire pitch here.
Try the free tools first
No account needed. Each one runs a piece of the same engine that powers Voxly.
A report on how you actually write on X: tone, formats, strengths, blind spots. Free.
Score your first line 0-100 before posting. Learn why 'I built a thing' underperforms.
Exact weighted counting via the official twitter-text parser. Runs fully client-side.
You could keep pasting into ChatGPT. Here's the honest comparison for this workflow.
Frequently asked
Will it keep the technical details accurate?
Drafts are grounded in the material you provide, and the chat makes targeted edits rather than regenerating, so your numbers and specifics survive editing. You review everything before publishing, and pinning a claim is as easy as selecting it and saying 'keep the exact figure'.
My timeline is mostly memes and replies. Can it still learn my voice?
The analyzer works from your original posts; if there aren't many, you can write the voice profile fields yourself in onboarding (it's a short, explicit form) and the system refines from what you publish going forward.
Can it turn a PR or changelog into a thread?
Yes, that's the flagship flow: paste the PR description or changelog, say 'thread, keep it dry', and refine per-tweet in chat. Long-form articles for bigger write-ups work the same way.
I don't want growth-hack stuff on my account.
Same. Voxly has no auto-DMs, no autoplugs, no engagement pods, and it never posts without your tap. It's a writer, not a growth suite; if you want automation machinery, other tools in the category do that and we compare against them honestly.
Is there an API or CLI?
Not yet. Today Voxly is the web editor plus the X API underneath. If you'd use a CLI or API for drafting from CI (changelog to draft on release), tell us; it's the kind of roadmap signal we weigh.
The post was already in the PR description.
Paste it, get your voice back, publish in one tap. Free plan, no card.