For most of the last year, the only way to tell us what to build into Shruvi was to email me. That worked when there were a few dozen of you. It does not scale, and more importantly, it hides the conversation. You never got to see what other people had asked for, or whether the thing you wanted was already on the roadmap, or whether it was already shipped two releases ago.
That changes today. voice.shruvi.com is now live.
What's there
Three tabs, all linked from the same nav.
Feedback. A public board where anyone can read posts, anyone signed in can submit one, and anyone signed in can upvote. Sort by top, new, or trending. The posts that get the most upvotes are the ones I look at first when planning the next release.
Roadmap. Four columns — Backlog, Next up, In Progress, Done. Public, read-only. When something moves from Next up to In Progress, you'll see it here before I write a release note. When it moves to Done, the release note is one click away.
Changelog. Every Shruvi release gets a markdown entry under
/changelog. Versioned URLs, RSS feed, the works. The five most recent
versions are already there, going back to v1.3.83. From now on, every tag
push that produces a desktop build also publishes a changelog entry — that's
a single source of truth instead of a Featurebase workspace, a GitHub
release page, and a forgotten Google Doc.
The voice part
Here is the feature I am most excited about.
On the new-post page, you'll see a microphone button next to the usual text input. Hold it, talk for up to sixty seconds, release. The audio is sent to the same backend Shruvi itself uses to transcribe your dictation. The transcript comes back, you review it, you edit it if you want, and you submit. The post shows up with a small mic icon next to the title so other people can see it was dictated.
The audio bytes are never stored. We transcribe in memory and discard the buffer the second the transcript is back. Only the text of your post lives on. This is the same privacy model Shruvi has always had for dictation, just applied to one more surface.
Why bother? Two reasons.
The first is that a dictation app whose own feedback board cannot be used by voice would be embarrassing. We dogfood our product. If voice.shruvi.com does not feel as natural to use by voice as the desktop app does, that is a bug we want to know about, fast.
The second is that I want feedback from people who would not otherwise sit down to type a paragraph. The bar is lower when you can hold a button and just say what's on your mind for thirty seconds. That's the entire pitch of Shruvi as a desktop app — speaking is faster than typing for most kinds of thought — and there is no reason it should not also apply to telling us what to build.
How to use it
Read whatever is interesting at voice.shruvi.com. No login required to read.
When you want to post or upvote, sign in with your existing Shruvi account.
The cookie is shared across *.shruvi.com, so once you're signed in on the
landing page or in the desktop app, you're signed in here too.
If you find a bug in voice.shruvi.com itself — including the voice recording flow — please post it on the board. I'd rather you used the thing to break the thing than email me about it.
The blog will keep going at blog.shruvi.com. Subscribe to the RSS feed if that's your thing. The next post is going to be about how I dictate code in Cursor without it autocompleting over my voice — something I have been asked about more times than any other single question.
Thanks for using Shruvi. Now go vote on something.
— Harish