Question 1

What is Vectora?

Accepted Answer

Vectora is a self-hosted AI agent with RAG, a native Context Graph, and multi-user web chat — it runs on your own computer or server, never in a Vectora cloud.

Question 2

How does self-hosting work?

Accepted Answer

You download and install Vectora (Windows, macOS, or Linux) — it's a native app, not a Docker image. Teams that need shared web chat usually run the Linux build on their own VPS; Docker is only used to spin up the optional infrastructure for 'complete' storage mode (Postgres+Qdrant+Redis), never for Vectora itself.

Question 3

Do I need a GPU to run it?

Accepted Answer

No. Vectora uses LLMs and embeddings via API (Google, Anthropic, OpenAI, Cohere, Voyage, OpenRouter) or local Ollama. For individual use, any recent laptop runs it fine; for a team VPS, 2 vCPUs and 4 GB RAM are already enough.

Question 4

Which operating systems are supported?

Accepted Answer

Windows 10+, macOS 12+, and Linux (Ubuntu 20.04+/Debian 11+/Fedora 35+) — native installers for all three, with built-in auto-update.

Question 5

How do I deploy on a server for my team?

Accepted Answer

Any Linux VPS with 2+ vCPUs and 4+ GB RAM — AWS, GCP, Hetzner, DigitalOcean, or your provider of choice. It's the same Linux installer (.AppImage/.deb/.rpm), just running on a server instead of your desktop.

Question 6

How long does installation take?

Accepted Answer

Less than 5 minutes: download the installer for your system, open the app, and create your root account on first launch. 'Complete' storage mode (optional, for scaling beyond the default SQLite+LanceDB) is deployed via a separate docker-compose — that part does take a few extra minutes.

Question 7

What's the difference between Free and Pro?

Accepted Answer

Free is free forever and requires no account: operational CLI, MCP, Desktop, and unlimited RAG (via Cohere/Voyage/Ollama), all running on your machine. Pro (R$24/month) requires an account and adds what only makes sense for teams: multi-user web chat, unlimited member invites, SSO/SAML, scalable storage (Postgres/Qdrant/Redis), webhooks, higher REST API rate limits, and priority support.

Question 8

Do I need to create an account to use Vectora?

Accepted Answer

No, not for local/solo use. The Free plan works 100% without login. An account is only needed for Pro's team features.

Question 9

Can I cancel anytime?

Accepted Answer

Yes. No minimum commitment. Cancel from the dashboard anytime — your account automatically reverts to the Free plan.

Question 10

Is my data sent to Vectora?

Accepted Answer

No. Everything stays on your server. Vectora has no access to your documents, conversations, or code.

Question 11

Is Vectora LGPD and GDPR compliant?

Accepted Answer

Yes. Since your data stays on your own infrastructure, you keep full control over compliance.

Question 12

How does authentication work?

Accepted Answer

On Free, it's local use with no login. For Pro (account required), Vectora's own authentication supports email/password and magic links.

Question 13

Which LLMs are supported?

Accepted Answer

Google Gemini (default), Anthropic, OpenAI, Cohere, Ollama (local), OpenRouter (proxy with hundreds of models), and any OpenAI-API-compatible provider.

Question 14

How does RAG work?

Accepted Answer

Embeddings via Cohere, Voyage, Ollama, or OpenRouter (local or hosted), LanceDB as the default vector store with dense vector search and optional reranking (Cohere/Voyage). In Complete mode (self-hosted with Postgres + Qdrant), search becomes hybrid (dense + BM25).

Question 15

Is there a public API?

Accepted Answer

Yes. REST API /v1 with per-plan limits (10 req/min on Free, 100 req/min on Pro). Docs at docs.vectora.company.

FAQ — Vectora

General

Installation

Plans

Security

Technical