Vectora
Vectora
Shared workspaceSelf-hostedPrivacy

Your workspace. Your AI.
Your server.

You and the agent work side by side in the same filesystem, terminal, git, and browser — not an isolated chat running hidden commands. Self-hosted: your data never leaves your server.

Vectora AI agent in action

See Vectora in action

Preview coming soon

Chat + code, in one place

Ask in natural language and watch the agent write, refactor, and explain code — all in the same conversation, with your repository's context.

Preview coming soon

RAG + Library

Semantic search over any document — plus a library of ready-made buckets: RAG, Skills, and community-published MCP connectors.

Preview coming soon

Reasoning + Remember

Complex tasks broken down automatically, step by step in real time — and Remember learns from your sessions, creating and updating skills on its own.

Preview coming soon

Kanban + subagents

Multi-agent board with background tasks — each subagent configurable as its own profile, role, and instructions, in an agent-company format.

How Vectora thinks

  • 1Orchestrator decides in real time: respond, delegate or parallelize
  • 2Coder Agent: files, terminal, git, code implementation
  • 3Search Agent: real-time web, RAG, knowledge base curation
  • 4RAG tools: vector search and optional reranking, available to the Search Agent
  • 5Parallel mode: independent tasks run at the same time
Technical docs →

A workspace, not a chat

Every capability below is shared between you and the agent — not hidden behind a tool call.

Browser

Real DevTools — console, network, a visual panel — plus Playwright and auto-detected dev servers. You and the agent see the same page.

Sandbox

Native isolation on every platform: bubblewrap + seccomp + Landlock on Linux, WSL2 on Windows, Seatbelt on macOS — not a generic Docker container.

Context Graph

A typed code graph across ~30 languages (tree-sitter), impact analysis, and GraphRAG — the same source of truth the agent's tool and your UI both query.

Terminal

A persistent PTY session — survives navigation and remounts, the same shell seen live by you and the agent, inside the sandbox, when active.

Git

Isolated worktrees, stash, diff/merge in an overlay, and PR creation without leaving the app.

Explorer

Optimistic conflict resolution when you and the agent edit the same file, full-text search shared with the agent's own tool, and per-commit file history.

Your documents, accessible anywhere

  • Install ready-made buckets from the Library — start already indexed, no waiting for the worker to process from scratch
  • PDF, DOCX, TXT, Markdown, and source code
  • Embeddings via Cohere, Voyage, Ollama, or OpenRouter — local or hosted
  • Dense vector search by default; hybrid search with BM25 in Complete mode (Postgres + Qdrant)
  • Optional reranking via Cohere or Voyage before responding
  • Background indexing — the embedding worker processes without blocking chat
  • Automatic curator summarizes the indexed base in the workspace MANIFEST
  • Source citation in every response
Pro plan

From zero to team in minutes

1/4

Deploy the stack

A single file. No external dependencies.

PostgreSQLQdrantRedis
2/4

Root account

Full administrative access to your workspace.

3/4

Invite team

Permission control per project.

4/4

Initialize projects

Each project has its own knowledge base and history.

Compatible with any Linux VPS — AWS, GCP, Hetzner, DigitalOcean

Also included in Pro

  • More efficient complete storage — hybrid search (BM25 + vector) via Qdrant, Redis caching, no manual tuning
  • SSO/SAML and unlimited member invites, with per-project permission control
  • Webhooks — trigger agent automations from GitHub, Slack, Linear, and other events
  • Connect — bring the agent to Telegram, Discord, Slack, or Email, no app switching

Why self-hosted?

Self-hosting isn't complexity. It's competitive advantage.

Privacy

No conversation, document or code is sent to third parties. LGPD, GDPR and internal policy compliance.

Cost

Fixed price for the Vectora license. LLM costs under your control — use local models for free.

Customization

Configure LLM providers, embedding models, chunk size, system prompts and much more.

Sovereignty

No lock-in. No cloud dependency. Run offline if needed. Your server, your rules.

Simple, transparent pricing

Free forever, no account required. Upgrade to Pro when you need team features.

Free forever
$0
  • Operational CLI + Desktop
  • RAG, Library, Browser, Sandbox, and Context Graph — all included
  • Model routing: Ollama, OpenRouter, 9Router, or connect directly (Gemini, Cohere, OpenAI, Anthropic)
  • 100% local, no account required
Download free
For teams and companies
$9/month
  • Everything in Free
  • Multi-user web chat
  • Unlimited member invites
  • SSO / SAML
  • Scalable storage (Postgres/Qdrant/Redis)
  • Webhooks
  • Connect
  • Priority support (SLA 24h)
Subscribe to Pro