Meet Tandi: ask your bookmarks, not your memory

A bookmark library can hold years of useful material and still leave you asking the same question: Where did I save that? Search helps when you remember the right title or keyword. Tandi is for the moments when you remember the question instead.

Tandi is BookmarkNa’s AI research assistant for your saved knowledge. Ask in everyday language and Tandi finds relevant bookmarks, summarizes what they say, surfaces useful connections, and keeps the supporting sources attached to the answer.

What is Tandi?

Tandi is the conversational layer over your BookmarkNa library. It works with the details you have already chosen to save—including titles, URLs, descriptions, notes, tags, folders, favorites, and reading activity—to help you retrieve and understand your own collection.

It is not a general chatbot searching the open web. Tandi begins with your library and answers in the context of the resources you collected. That narrower scope is intentional: the goal is to make your saved knowledge more useful, not add another stream of unrelated information.

Questions Tandi can answer

You do not need to learn special commands. Start with the outcome you want or the part of your library you are trying to understand.

Tandi also adapts its suggested questions to the folders, tags, notes, and code present in your library. The starting ideas become more useful as your collection develops.

Answers grounded in your library

An AI summary is only useful when you can verify where it came from. Tandi places the relevant BookmarkNa resources below each answer, including their saved notes and tags, so you can inspect the original material and continue working.

• Source bookmarks show which saved resources support the response.
• Key takeaways separate the most useful findings from the longer summary.
• Caveats appear when the available library context is incomplete or the answer needs interpretation.

This source-first design keeps Tandi connected to retrieval. The answer is a useful starting point, while your saved material remains the record you can open, compare, and judge.

Simple lookup locally, deeper synthesis with AI

Not every question needs an AI model. Tandi first determines what kind of work the question requires. Straightforward requests—such as recent bookmarks, favorites, counts, tags, or direct matches—can be answered with local bookmark logic and do not use AI credits.

Questions that ask for broader themes, comparisons, or synthesis can use Tandi’s AI service when it is configured. If that service is unavailable, Tandi falls back to local bookmark matches instead of leaving you with an empty result.

Private library, visible sources, user control

Tandi is available inside your signed-in BookmarkNa workspace. Requests are scoped to the account making them, and the answer returns only the bookmarks selected from that user’s library.

You decide what enters the library, what question to ask, and which source to open. Tandi does not silently browse external websites or change your collection in the background. Review BookmarkNa’s privacy policy for broader details about account and service data.

Start asking Tandi

1. Open your BookmarkNa workspace.
2. Select Ask Tandi from the sidebar.
3. Choose a suggested question or write your own.
4. Review the summary, key takeaways, and any caveat.
5. Open the source bookmarks that support the answer.

If your library is still small, begin with direct prompts about recent saves, favorites, or a specific tag. Richer summaries become possible as you add useful titles, descriptions, notes, and organization.

What Tandi does not do yet

Tandi currently helps you find, summarize, and understand saved resources. It does not autonomously edit titles, move bookmarks, create folders, apply tags, archive items, or delete content. Those actions remain under your direct control.

We are deliberately starting with trustworthy retrieval before expanding into agent actions. Future capabilities should make organization easier without taking irreversible steps or obscuring what changed.