Focused by Design: How Vault-Scoped Chat Makes Your AI Smarter

By Finn

Executive Summary

Vault-scoped chat lets you confine Zibri.ai's AI chat to a single vault, so every answer draws only from the notes, documents, and recordings you've grouped for one project or topic. The result is more relevant responses, cleaner source attribution, and a smarter use of your monthly AI query budget. If you're already organizing your work into vaults, this feature turns each one into its own dedicated AI assistant.

Introduction: What Is Vault-Scoped Chat?

Zibri.ai's AI chat is built on a straightforward idea: your notes become your AI. Rather than pulling from generic internet data, it draws from your own research, your own words, your own proprietary knowledge. That's the core value proposition.

Vault-scoped chat takes that idea one step further. Instead of querying your entire knowledge base, you can scope a chat session to a specific vault — and every answer the AI generates will come exclusively from the content inside that vault.

"Scoping chats to a specific vault to get focused responses on one project or topic area." — Zibri.ai Documentation

If you have a vault for a client engagement, a research project, or a writing project, vault-scoped chat turns that vault into a targeted assistant that only knows what you've put into it. That's a meaningful shift from a general-purpose query tool to something that feels purpose-built for the work in front of you.

Why Scope Your Chat? The Problem with Broad Queries

When you query your entire knowledge base, the AI has a lot to work with — and that's not always a good thing.

Zibri.ai uses Retrieval-Augmented Generation (RAG), which means the AI retrieves the most relevant content from your knowledge base before generating an answer. Each response shows which notes or documents it drew from. That source attribution is useful. But when your knowledge base spans multiple projects, clients, and topics, "most relevant" becomes harder to define. The AI might surface a note from a different project that shares a keyword, or pull context from an old document that's no longer current. The answer is technically grounded in your content — but not necessarily in the right content.

There's also a practical consideration: query limits. The Personal tier includes 500 AI queries per month. The Pro tier includes 3,000. Those aren't unlimited. A broad query that returns noisy, partially relevant answers doesn't just waste a question — it may send you back to ask a follow-up, burning another query to clarify what a more focused first query would have answered cleanly.

Scoping your chat to a vault solves both problems. It narrows the retrieval pool to content that's actually relevant to the task at hand, and it makes each query count.

How Vault-Scoped Chat Works in Zibri.ai

The mechanics are worth understanding, even at a high level, because they explain why vault-scoping produces better answers — not just different ones.

Zibri.ai's AI chat uses RAG: when you ask a question, the system retrieves the most relevant content from your knowledge base and uses it to generate a sourced answer. Each response shows which notes or documents it drew from. Zibri doesn't hallucinate facts — the answers are grounded in your actual content.

When you scope a chat to a vault, you're telling the retrieval step to look only inside that vault. The AI doesn't have access to notes from your other vaults during that session. The retrieval pool shrinks, and the signal-to-noise ratio improves. You get answers that are tighter, more specific, and easier to trace back to their sources.

You can also layer in a custom system prompt. Custom prompts let you adjust the AI's tone and response style within the chat settings — for example, asking it to respond in bullet points, use a formal register, or summarize rather than explain. Vault-scoping and custom prompts work together. One controls what the AI draws from; the other controls how it presents what it finds.

Step-by-Step: Scoping a Chat to a Vault

Vault-scoped chat is a customization option within Zibri.ai's chat settings. Here's how to activate it:

1. Open AI Chat. Navigate to the chat interface in Zibri.ai.

2. Open chat settings. Look for the settings or configuration option within the chat panel.

3. Select your vault. Choose the specific vault you want the AI to draw from. Only content inside that vault will be used to generate answers during this session.

4. Optionally, add a custom system prompt. If you want the AI to respond in a particular format or tone — bullet points, formal language, brief summaries — enter your custom prompt in the system prompt field within chat settings.

5. Ask your question. Type your query as you normally would. The AI will retrieve content from the selected vault and generate a sourced answer.

6. Review the sources. Each response shows which notes or documents it drew from. Use those citations to verify the answer and trace it back to your original content.

That's the full workflow. The vault selection is the key step; everything else is optional refinement.

Use Cases for Vault-Scoped Chat

The feature is most useful when your knowledge base covers multiple distinct areas and you need answers that stay in one lane. Here are a few scenarios where vault-scoped chat earns its place.

Project-specific research. If you're a researcher with separate vaults for different studies or literature reviews, scoping your chat to one vault keeps your AI queries focused on that body of work. You won't get answers contaminated by notes from a different project.

Client-specific consulting. Consultants often maintain separate vaults for each client — meeting notes, documents, voice recordings, background research. Vault-scoped chat lets you query a single client's knowledge base without pulling in context from other engagements. That matters for accuracy and, frankly, for keeping client information appropriately separated.

Writing and drafting. Writers who organize their research by project can use vault-scoped chat to ask questions about a specific manuscript's source material. "What did my interview notes say about the timeline?" becomes a precise query when the AI is only looking at the vault for that project.

Topic deep-dives. If you maintain a vault on a specific subject — a technology area, a market, a discipline — vault-scoped chat lets you interrogate that vault directly. You're not asking what you know generally; you're asking what you've captured on this topic specifically.

In each case, the underlying principle is the same: your knowledge is more useful when it's organized, and vault-scoped chat makes that organization actionable.

Tips for Getting the Most Out of Vault-Scoped Chat

A few practices will help you get sharper answers and protect your query budget.

Keep vaults focused. A vault that contains everything about a project — notes, documents, voice recordings — is more useful than one that mixes unrelated content. The cleaner the vault, the more precise the retrieval.

Pair vault-scoping with a custom system prompt. If you always want answers in a specific format for a particular project, save that prompt alongside your vault workflow. A prompt like "respond in three bullet points, citing the source note for each" can make your outputs immediately usable.

Watch your query usage. Zibri.ai notifies you when you've used 80% of your monthly query allowance. Vault-scoped chat helps you direct those queries toward the content that matters most, rather than spending them on broad searches that return partially relevant results.

Ask follow-up questions within the same scoped session. Once you've scoped a chat to a vault, stay in that session for related questions. You'll maintain the focused context and get more coherent, connected answers across a line of inquiry.

Conclusion

Vault-scoped chat is a small configuration choice with a meaningful practical effect. It turns a broad AI assistant into a focused one — answering questions from exactly the knowledge you've curated for a specific project, client, or topic, and nothing else.

The broader Zibri.ai premise is that your notes should become your AI, trained on your research and your proprietary knowledge rather than generic data. Vault-scoped chat is where that premise becomes most concrete. Each vault you build can function as its own targeted assistant, drawing on the specific content you've put into it.

If you're already using vaults to organize your work, the next step is straightforward: scope your next chat session to one of them, and see what a focused query actually feels like.