A Self-Hosted Research Workspace with Cited Answers
Arivu keeps web sources, documents, transcripts, notes, research objects, tasks, reminders, and cited answers in one self-hosted SQLite-backed workspace.
Research tools often split your work into pieces.
Sources live in a bookmark manager. Documents live in a folder. Notes live in a note app. Tasks live in a task manager. Reminders live in a calendar. Answers come from a chatbot that may know nothing about what you saved. Exports exist, but only if you remember to run them before you need them.
That fragmentation creates friction. You lose the thread between source, thought, action, and answer.
Arivu’s current direction is to keep those pieces together in a self-hosted workspace.
Save sources where they can become useful
Arivu captures web pages from the dashboard, browser extension, installable app share target, or CLI. A capture can include quick notes, tags, and selected quotes. The server fetches and sanitizes page content, then stores the saved item in your own SQLite-backed instance.
That alone is not enough. A source needs context.
In Arivu, a saved page can carry annotations, linked notes, explicit links to other items, tasks, reminders, review state, and source metadata. A note can do the same. Documents and transcripts can become notes too, so imported PDFs, EPUBs, Markdown files, HTML files, transcript text, and OCR text join the same workspace instead of becoming a parallel archive.
Ask questions against your own library
Generic AI answers are useful for exploration. They are less useful when you need to know what your own sources say.
Arivu’s cited-answer mode works from saved content. It uses your summaries, highlights, snippets, linked context, and notes to synthesize an answer, then points back to the source items it used.
That changes the trust model. You are not asking the open web to improvise. You are asking your saved library to explain itself.
For example:
- “What did I save about SQLite backups?”
- “Which sources mention review loops?”
- “Summarize my notes about migration planning.”
- “What are the arguments against adding another dependency?”
The answer is useful because it is grounded in material you chose to keep.
Keep actions attached to sources
Research often creates work:
- read this before the meeting
- extract quotes for a post
- compare this with another source
- revisit this after the launch
- send this note to a collaborator
If those actions move into a separate task app, the context starts to decay. The task says “review article,” but the reason is gone.
Arivu keeps tasks and reminders attached to bookmarks, imported document notes, and standalone notes. Focus gathers those open loops by pending, overdue, today, upcoming, and completed views. Review can bring stale tasks and due reminders back when they need attention.
The source and the commitment stay together.
Use notes and objects for research structure
Standalone notes are first-class in Arivu. They are not just comments on bookmarks.
Use them for:
- a daily research log
- a synthesis note
- an argument draft
- a reading question
- a meeting prep note
- a decision record
Notes can link to bookmarks and other notes. Backlinks show what points at them. Tasks and reminders can attach directly to the note. This makes a note a working object, not a dead text field.
When a note becomes a recurring thing, turn it into a research object. Arivu supports lightweight objects for projects, people, books, meetings, decisions, and research threads. A decision can preserve rationale. A meeting can connect to source notes. A research thread can follow a question across daily notes, saved pages, imported documents, and later decisions.
Topic evolution helps with that last part. It lets you trace a topic across bookmarks, notes, daily notes, meetings, decisions, and objects, so you can see how the work changed instead of only retrieving isolated hits.
Import, export, and leave cleanly
Research work needs a way in and a way out.
Arivu can import common bookmark exports, browser HTML, Arivu JSON backups, URL lists, OPML, RSS/Atom, URL-bearing CSV/TSV formats such as Readwise or Kindle-derived tables, documents, transcripts, OCR text, and calendar events. Import jobs show progress, source reports, fetched counts, AI-processed counts, failed counts, and a bounded sample of imported items.
Exports include JSON backup, CSV, browser HTML, Markdown, and Obsidian-ready ZIP output. The JSON backup is meant for restore and includes notes, links, annotations, review history, and research objects. The Markdown and Obsidian outputs are meant for portability.
If a second brain cannot leave, it is not really yours.
AI helps, but the workspace still works without it
Provider integrations are optional. Connect Gemini and Arivu can use AI-powered processing where available. Leave providers unconfigured and the workspace still works: capture, notes, research objects, Inbox, Focus, Review, tasks, reminders, imports, exports, search, and local deterministic enrichment remain useful.
That matters for private research. Sometimes you want external AI. Sometimes you want a local, self-contained workflow. A self-hosted research workspace should let you choose.
Who this is for
Arivu is for people who save and revisit web material as part of their work:
- researchers managing sources, documents, and transcripts
- writers collecting references
- operators documenting decisions
- developers saving technical context
- founders tracking market and product material
- lifelong learners who want recall, not just storage
It is not trying to replace every note app, task app, or reading app. It is trying to make saved web knowledge usable: captured, connected, actionable, reviewable, answerable, and portable.
That is the difference between a bookmark archive and a research workspace.