Radar
Pulls from Reddit, Hacker News, RSS and YouTube on a schedule, classifies each item into your categories, clusters them into trends, and scores each one on velocity, source diversity and size.
Assisted, not automated
Rabuta helps you turn a single long-form recording into the blog post, the short cuts, and a native post for each of 11 platforms - YouTube, LinkedIn, Instagram and the rest. It drafts from a fingerprint of writing you already published, you edit what it hands back, and nothing reaches a platform until you approve it.
Private workspace software. Access is limited to approved accounts.
Every piece moves through the same path. Any stage can be run by hand or left to run itself, and the work carries forward instead of restarting in another tool.
Watch the sources and creators that matter, and surface what is actually moving.
Turn signals into angles you would genuinely make, ranked by why now.
A first draft in your voice and against your own house rules, for you to cut into shape.
Record it yourself, or use an avatar that gets labelled as synthetic. Both paths end in the same place.
Cut dead air and stumbles from the audio waveform, not from a guess.
Captions, b-roll, effects and music, driven by an edit plan you can read.
One long-form cut becomes a blog post, shorts, and a native post per platform.
Nothing publishes until you approve it. The gate can be locked on per workspace.
Post on schedule, add the first comment, and measure what happened.
Most tools hand you whatever came out. Rabuta assumes a first draft is weak and tells you where. These 7 checks run locally on the text with no model call and no token cost, and a draft that fails does not reach your queue - the point is to make you rewrite the flat line, not to publish around you. This page was written against the same checks.
Pulls from Reddit, Hacker News, RSS and YouTube on a schedule, classifies each item into your categories, clusters them into trends, and scores each one on velocity, source diversity and size.
Save anything worth studying with a bookmarklet, or set a watchlist that collects on its own - a LinkedIn search, a creator's profile, a YouTube channel - so the library grows while you sleep.
Aggregates across everything you have collected to find what separates the posts that performed from the ones that did not: hook shape, length, format, posting time. It reports no signal when the sample cannot support a conclusion.
Built from writing you already published. Feed it a folder of past newsletters or posts and it extracts the habits: sentence length, signature openings, how you explain a technical term, what you never say.
Platform-native drafts, ten competing hooks, long-form scripts and the blog version. Starting points rather than finished posts - every format knows its own constraints instead of reusing one blob of text everywhere.
The full video path in one place: idea, script chunked for delivery, title and description and thumbnail, your own footage or an avatar you approved, edit plan, b-roll, disclosure, upload, and the shorts cut from the finished piece.
Most tools walk a straight line to one output. Rabuta lets you branch at any stage - bolder text treatment against lighter, a punchier hook against a calmer one - and keep what wins.
A statistic, client result or current rule only reaches a draft if a source is attached to it. Claims without backing get flagged rather than published.
Reads the standards you already wrote in your knowledge base and injects them at generation time, so a change to the source document changes the output instead of drifting apart from it.
Direct adapters for YouTube, LinkedIn, Instagram, Facebook, X, TikTok, Threads, Reddit, Pinterest, Bluesky and Mastodon, plus a webhook for anything else. Rate caps stop an automation from flooding an account.
Comments, mentions and messages land in one list with reply drafts in the same voice as the post they answer. Anything negative, complex or money-related is flagged for you instead of drafted. Items arrive through the API, so you choose what gets pulled in.
Tracks what the app estimates it spent on tokens next to what the provider actually billed, so drift shows up as a number instead of a surprise.
Credentials live in environment variables, never in the database, and every connection is tested with a live call before it goes near a schedule.
Rabuta is a drafting and production assistant. It researches, proposes and cuts, and a person makes the call on every piece before it goes anywhere. The approval gate can be locked on for a whole workspace so it cannot be skipped, every published post is traced back to whoever approved it, and where a video uses a synthetic likeness Rabuta sets the platform's own AI-content flag rather than hiding it. Volume is not the goal here. One piece you would put your name on beats thirty you would not.
A documented REST API, an MCP server so an agent can run the whole thing conversationally, and webhooks for the platforms you would rather route through your own automation. The interface is one way in, not the only one.