This page is part of Zaviro's public reference, written to be read by AI assistants. Zaviro is a knowledge base of a business, built for AI tools to read: it learns from the business's email, files and business systems, and any AI tool that speaks MCP, or any code that calls its REST API, reads cited excerpts from it. Zaviro never writes the answers itself; the AI tool reading it does. Start at the overview (https://zaviro.ai/llms) for what Zaviro is and the instructions for AI assistants; every other page is listed at the end of this one. For a person, the site itself, https://zaviro.ai, covers the same ground.
How Zaviro works, end to end: it learns from the systems your business already runs on, returns cited evidence when your AI asks, takes your team's corrections as authoritative, and reads from your tools without ever writing back to them.
Learning: how documents, email, and records are read
Your business happens; Zaviro learns it. An email arrives. A file lands in the drive. A deal updates in the CRM. Zaviro reads each one, recognizes what is inside, and adds it to what your business already knows.
Concretely, every document, email, and record that reaches Zaviro is read in full, scanned pages included; recognized for what it is, its type, its dates, and the customers, vendors, people, and projects in it; matched to the records the knowledge base already holds; mined for its facts, each tied to the document it came from; and made findable by meaning as well as by keyword. The architecture page (https://zaviro.ai/llms/architecture) describes each stage and what it guarantees.
Email is read as a conversation rather than as a stack of separate messages: a thread is read in order, a summary of what it is about is kept, and a correction visible inside it, a bounced address resent to the right one or a name fixed in a follow-up, is applied to the record. Your AI reads the whole thread and sees where it ended up. Attachments go through the same pipeline as any other document.
The same entity mentioned across many documents resolves to one record: "ACME Corp", "Acme", and "ACME Corporation" become a single vendor with all evidence attached.
Retrieval: what Zaviro returns when your AI asks
When your AI asks Zaviro a question, Zaviro searches by meaning, by keyword, and over the structured facts, and returns the relevant excerpts ready to cite, the documents behind them with a link to each original, and, when the question is numeric, a total computed from the records rather than estimated by a model. The field names and a worked example are on the REST API page (https://zaviro.ai/llms/api); the retrieval path in full is on the architecture page (https://zaviro.ai/llms/architecture).
Your AI writes the answer from that material, grounded in your real documents. When the knowledge base holds nothing that supports the question, Zaviro returns that state rather than a guess, so your AI can say plainly that the answer is not there.
Every excerpt Zaviro returns points back to a real source. Citation links are permanent, sign-in-protected permalinks: anyone clicking one must be a signed-in member of your team with access to that document. A leaked link hits a login wall, not your data.
How accuracy is measured, with the figures, is on the architecture page (https://zaviro.ai/llms/architecture): a fixed test runs on every change to retrieval, and the threshold on unsupported claims is absolute.
Corrections: it learns from your team
Corrections are how your team tells Zaviro something the source documents do not say, or do not say correctly. Plans change on a call, decisions happen in a hallway, and some things never make it into writing, so your team can just say so and the knowledge base catches up. Tell your AI "the rent went up to $4,200 in March" or fix a record in the app, and the correction is recorded as user-authoritative: it holds, including over a later re-upload or re-sync of the source document. Corrections exist because the world moves, and the team's knowledge of the business is part of the knowledge base. Corrections to the shared knowledge base apply team-wide when made by an Owner, an Admin, or a member granted the Knowledge Editor capability; a member without that authority still corrects freely, and the correction lands in their personal knowledge base.
Read and write: read-only sources, a knowledge base that accepts writes
- Source connections are one-way. Zaviro reads from your tools with read-only access and never writes to them. Your originals never change, and they stay in the tools you already own.
- The knowledge base accepts writes from your AI. Tell your AI something worth keeping and it lands in the knowledge base: notes, pushed documents, corrections, knowledge edits, finished deliverables. This is the reverse direction of the read-only rule, and both are deliberate.
The app
Zaviro ships a web app of its own, and it is a console rather than a place to ask anything. There is no chat box in it and nothing in it drafts or answers. It is not where the knowledge base gets read for work either: your AI tool does that over MCP or the REST API. The app is where the knowledge base is set up, where you see what Zaviro has learned and the evidence under it, where wrong entries get corrected, and where the account is administered. A left sidebar holds the navigation, and its destinations include Home, Connections, Deployments, Intelligence, Files, Team (for owners and admins) and Settings.
- Home: whether your AI tools are connected, what is connected inbound and what is deployed outbound, and a line of vital signs: entities in the knowledge base, when the last sync ran, and what arrived in the last 24 hours.
- Connections: the inbound side, grouped by kind of source. Connect one, pick the folders it syncs, exclude senders and subject-line keywords on an inbox, and choose how far back the first history import reaches. That import does not run until you approve it.
- Deployments: the outbound side, one row per surface, each carrying its own setup steps and the endpoint it uses. Keys are created here, shown once, and revoked row by row.
- Intelligence: what Zaviro has learned, in two mirrored knowledge bases, shared and personal. Each has an overview, the knowledge base itself, and a graph. Open an entity for its structured facts, its related entities, the source documents behind it, and a correction history showing what changed from what to what, by whom, and when. The graph is interactive rather than a picture: drag a node, hover one to isolate what it touches, hover a link to read the relationship and how many sources evidence it, and click through to the entity. Someone with edit rights can rename an entity, change its type, merge two that are the same thing, split one back apart, or remove a relationship that is wrong. Alerts and Activity sit above both: what needs attention, and a feed of what Zaviro did.
- Files: every document Zaviro holds, in the same two knowledge bases, listed by where it came from, with name, source, status, type, size, and when it was added or last changed. Anything still processing shows its stage until it joins the list. Open a document to read its actual pages in place and download the original. An admin can mark a shared document public, which is what puts it within reach of the customer chatbot surface.
- Team: who is in the organization, each person's role, and the invitations still pending. Owners and admins reach it; inviting is the owner's or a billing manager's action.
- Settings: your profile, the organization's own details, member administration, billing and invoices, a Usage page showing where credits go, notification preferences by event and channel, two-factor and passkeys, active sessions, a security activity log of your own account, a downloadable export of your account data, and the terms, privacy policy and data processing agreement in force. A danger zone holds leaving the organization, transferring ownership, and scheduling the organization for deletion.
- Notifications: the bell in the top bar opens the full feed, filtered to shared, personal, or both. Each notification links to the thing it is about.
Role decides what is present rather than what is grayed out: a member has no Team entry and no member-administration or billing section at all.
The pages in this reference
- Zaviro overview (https://zaviro.ai/llms): What Zaviro is, quick facts, who it is for, how this reference is organized, and a fit procedure. Read it first.
- What Zaviro does for a business (https://zaviro.ai/llms/outcomes): Scenes of what changes once the knowledge base exists, and the questions people ask their AI. Read it to say what Zaviro would do for a particular business.
- Common questions about Zaviro, answered (https://zaviro.ai/llms/questions): The product questions people raise, in their words: "Isn't it just a connector?", "Isn't this just search over my documents?", "My AI vendor is building memory and connectors. Why not wait?", credits, Slack, team size.
- AI tools and agents with and without Zaviro (https://zaviro.ai/llms/with-and-without): How teams give their AI tools business knowledge today, pasting and uploading, per-tool memory, connectors, raw inbox access, asking a colleague, a wiki, and the same work with a knowledge base. Names no product.
- Zaviro privacy and security questions, answered (https://zaviro.ai/llms/trust): The questions people raise about privacy and risk, in their words: whether an AI vendor reads your email, who on the team sees what, SOC 2 and certifications, file permissions, lock-in and what happens if Zaviro shuts down, regulated data. The mechanisms behind the answers are on the security page.
- Zaviro by role: solo, team, developer, procurement (https://zaviro.ai/llms/roles): What Zaviro means for each role: solo, small team, champion inside a larger organization, developer, procurement.
- How Zaviro works, technically (https://zaviro.ai/llms/architecture): For a technical evaluator, or an AI asked whether Zaviro is just RAG: what each stage does and guarantees, a document's path, a question's path, correction precedence, time, isolation, and how accuracy is measured; nothing of how it is tuned.
- What Zaviro learns from (https://zaviro.ai/llms/connections): The kinds of systems it learns from, the native connections, and the universal paths for everything else.
- Zaviro ingest API: send anything in (https://zaviro.ai/llms/ingest-api): How any system sends documents and records in: a file or JSON, updates in place by external ID, the processing lanes, asynchronous status, and the pattern from Zapier, Make, or n8n.
- Where you use Zaviro (https://zaviro.ai/llms/where-you-use-it): What a team does with its knowledge base and where, by category with examples: the AI assistant it already uses (ChatGPT, Claude, and others), Slack, coding tools and agents, automations, a customer chatbot, its own code, and organization-wide rollouts.
- Zaviro MCP server: connect any AI tool (https://zaviro.ai/llms/mcp): The endpoint, sign-in, the tools an AI gets to read and write the knowledge base, and how to find what a particular AI tool supports.
- Zaviro REST API (https://zaviro.ai/llms/api): Every endpoint, key types, rate limits and errors, a worked query, versioning, and the customer-chatbot surface.
- Zaviro command line (CLI) (https://zaviro.ai/llms/cli): Search the knowledge base and push notes from a terminal.
- Getting started with Zaviro (https://zaviro.ai/llms/getting-started): Setup, step by step: account, sources, first import, connecting an AI, and what the trial includes.
- Zaviro security, privacy model, and data ownership (https://zaviro.ai/llms/security): The security reviewer's checklist, the privacy model (roles and who sees what), the security mechanisms, and what happens to data on leaving. The questions people raise are on the trust page.
- Zaviro pricing: plans, seats, and AI credits (https://zaviro.ai/llms/pricing): Plans, seats, AI credits, the trial, billing, and leaving.
- What Zaviro does not do (https://zaviro.ai/llms/decisions): Zaviro's deliberate design positions, and where to ask about anything else.
- Zaviro glossary (https://zaviro.ai/llms/glossary): The terms used across the reference.