Using Ounie

Brain settings

Open a brain and choose Settings to control who can see it, who can help build it, how it answers, and how it leaves. Here is what every setting does.

The Settings tab has its own left sidebar: General, Widget, Suggestions, Research, Model, and Marketplace. General holds the everyday controls; the rest are focused panels you only need when you reach for them. Settings are owner-only — collaborators don't see this tab.

General#

Details#

The brain's identity. Set an icon (upload a square PNG or JPG up to 5 MB, or pick an emoji), a name, and an optional description. These show up everywhere the brain appears — your dashboard, public pages, and shared links. Click Save changes to apply.

Visibility#

Who can reach this brain at all. There are three levels:

• Private — only you can see this brain. This is the default; nothing leaves your account.
• Shared — only people you invite can access it. Add them under People below and give each one a role. The brain stays off the public web.
• Public— anyone with the link can read it. Pages get clean, search-indexed URLs and visitors can ask the brain questions. Several other settings (public questions, suggestions, forking's fork button, marketplace listings) only take effect once a brain is public.

People & roles#

Invite collaborators by email or @username and assign each one a role. You are always the Owner (shown with a crown). Roles are cumulative — each one can do everything the role below it can, plus more:

• Viewer — can read every page and ask the brain questions. Cannot add or change sources. The lightest level of access.
• Contributor — everything a Viewer can do, plus add new sources (files, links, notes). Their additions go straight into the brain, but they cannot edit or delete existing sources and pages.
• Editor— everything a Contributor can do, plus edit and delete existing sources and hand-edit wiki pages. Full control of the brain's knowledge, but not its settings.
• Owner — you. Everything an Editor can do, plus manage settings: visibility, people, billing-gated features, and deletion. There is one Owner per brain.

Forking#

Let anyone copy this brain's wiki into their own, with attribution. A fork copies the synthesized, readable layer and the graph — never your raw files. Forks start private and credit the original. Sharing covers the full picture.

Show questions publicly#

Let anyone see this brain's question history on its public page, and upvote or save useful questions. Answers are shown; the underlying chat is never editable by visitors. This only applies to public brains — make the brain public to enable it.

Export this brain#

Your knowledge is yours — download it anytime, on any plan. Choose a format:

• Obsidian vault — Markdown with [[links]] that opens straight into Obsidian.
• Markdown — plain .md files, one per page.
• JSON — a structured backup of pages, links, and the graph.

Two options refine the bundle: Include original files adds your uploaded PDFs and images under an _attachments folder (and more size), and Folder by kind groups notes, sources, concepts, and entities into folders. The export runs in the background; download it when it's ready. See Exporting for details.

Danger zone#

Delete brainremoves all of its sources, pages, and history. This can't be undone — you'll be asked to confirm. Consider exporting first.

Widget#

Configure the embeddable Ask bubble for this brain — the same one you drop on your website with a script tag. Here you create widgets and set each one's name, accent color and light/dark theme, greeting, suggested questions, allowed origins (the sites permitted to embed it), and a daily cap. You can also set a combined per-brain limit across all widgets as a guardrail on credit spend; each widget question costs 0.5 credits beyond your plan's included allowance. Full guide in Widgets.

Suggestions#

Let logged-in visitors suggest a link or note for a public brain. Nothing is added until you accept it — accepted suggestions run through your normal source pipeline and count against your source limit. This panel holds the Accept source suggestions toggle (available once the brain is public) and an inbox of pending, accepted, and rejected suggestions to review.

Research#

Auto-Research keeps a brain current automatically. On a cadence Ounie discovers fresh material from the web and RSS feeds you name, filters it for relevance and novelty, and adds it. You write a research brief, choose sources, freshness, cadence and how many items per run, and pick an approval mode: Review waits for your approval in an inbox (credits are spent only on approved items), or Auto adds relevant items on its own. Each added item costs 3 credits. This is a Pro feature. See Auto-Research.

Model#

Choose how this brain turns sources into its wiki. By default it uses the shared house model. You can instead bring your own:

• My API key — synthesize on your own provider (Anthropic, OpenAI, OpenRouter, NVIDIA, or any OpenAI-compatible endpoint). Your key is stored encrypted and pinged when you save. Optionally use it for answers too, for unlimited private asks; public, widget, and marketplace questions stay on the house model.
• External via MCP — new sources extract their text and wait for your own LLM client (Claude Code, Claude.ai, Cursor) to pull the raw text over MCP, synthesize it, and push the result back.
• House model — the shared Ounie model. The default; reset to this anytime to clear your settings.

Bring-your-own-model is a Pro feature. An account-level default applies to every brain, and each brain can override it here.

Marketplace#

Sell a public brain's answers for credits. Set a price per question, optionally offer a few free preview questions before the paywall, and write a blurb. Human askers pay in credits; autonomous agents can pay per call via x402. You earn credits on every ask and cash them out. To publish, the brain must be public and you need a payout method — until then you can save a draft. This tab appears only on plans that allow selling. See Marketplace.