Skip to content

Curiosity

What it is

Curiosity is a desktop-first AI search application and knowledge assistant that provides a unified interface for searching across local files, emails, and cloud storage. As of July 2026, it has expanded into the Curiosity Workspace platform, offering enhanced enterprise features, SSO support (OIDC/SAML), and deep integration with local LLMs (via Ollama) and multi-model vector indexing. - Licensing: Proprietary (Freemium) - Cost: Free (Personal) / Paid (Pro & Workspace) - Self-hostable: Desktop app (Local data) / Workspace (On-premise option)

What problem it solves

It solves the problem of "information fragmentation" where data is scattered across multiple SaaS apps (Slack, Jira, Notion) and local folders. Curiosity provides a single "source of truth" for search, combined with an AI assistant that reasons over indexed data locally, ensuring privacy and reducing the need to upload sensitive files to public clouds.

Where it fits in the stack

Enterprise AI / Personal Productivity / Desktop Search. It acts as a human-facing "Agentic Interface" that bridges the gap between local files and cloud-based knowledge.

Typical use cases

  • Unified Global Search: Finding a specific email attachment, Slack thread, or Jira ticket using a single global keyboard shortcut.
  • Private Local RAG: Asking questions about your local PDF library or code documentation using a local model via Ollama.
  • Workspace Collaboration: Grouping related files, notes, and emails into "Spaces" that can be shared across a team with centralized SSO.
  • Agentic Automation: Utilizing AI agents that can retrieve information, summarize threads, and even "ask" the user for clarification mid-task.

Strengths

  • Privacy-First Architecture: Most indexing and AI processing (with local LLMs) occur on the user's machine.
  • Native Desktop Experience: High-performance, keyboard-driven interface with instant "Launcher" access.
  • Extensive Connectors: Supports 50+ cloud and local sources including Microsoft 365, Google Workspace, GitHub, and Notion.
  • July 2026 Features: LLM Usage Dashboard (cost/token tracking), Multi-Model Vector Indexing (run embedding models side-by-side), and Agentic Questioning (human-in-the-loop support).
  • Advanced Filtering: Robust inline filters (e.g., @file, ext:, src:) for precision search.

Limitations

  • Closed Source: The core application and Workspace server are proprietary.
  • Resource Intensity: Indexing large datasets and running local LLMs can significantly impact system CPU and RAM.
  • Desktop Focus: While a web version exists for Workspace, the primary power and local indexing require the desktop agent.

When to use it

  • If you value privacy and want to search local files alongside cloud data without centralized storage.
  • If you find yourself constantly switching between browser tabs and local folders to find project info.
  • If you want a desktop-native AI assistant that "knows" your work history across multiple apps.

When not to use it

  • If you strictly require 100% open-source software (consider Khoj).
  • If you prefer a pure web-based experience and do not want to install a local agent.
  • For high-performance, cluster-wide enterprise search where a dedicated engine like Elasticsearch is required.

Getting started

Installation

Download the installer for your platform from curiosity.ai. - macOS: DMG or Homebrew Cask. - Windows: MSI/EXE. - Linux: AppImage, DEB, or RPM.

Connecting Local LLM (Ollama)

  1. Ensure Ollama is running on your machine.
  2. In Curiosity, navigate to Settings > AI Assistant.
  3. Select Local LLM (Ollama) as the provider.
  4. Choose your preferred model (e.g., gemma3:27b) and click Connect.

CLI examples

Curiosity Workspace includes a CLI for administrative tasks, and it supports the Model Context Protocol for agentic integration.

# Register Curiosity as an MCP server for an agent (July 2026)
mcp register curiosity-server --command "curiosity-mcp" --args "--workspace-url https://my-org.curiosity.ai"

# Trigger a re-index of a specific source via Workspace CLI
curiosity-cli index trigger --source "google-drive-shared" --workspace "enterprise-docs"

# Launcher Shortcuts (Keyboard-first productivity)
# Alt + Space (Win/Linux) or Cmd + Space (Mac): Toggle Launcher.
# / : Start a command or search filter (e.g., /type:pdf).

API examples

Curiosity Workspace provides a REST API for automated data ingestion and triggering AI tasks using models like Gemma 3 or Claude 5.1.

Triggering AI Tasks (Python)

import requests

API_TOKEN = "YOUR_WORKSPACE_TOKEN"
API_URL = "https://your-workspace.curiosity.ai/api/v1/tasks/summarize"

headers = {
    "Authorization": f"Bearer {API_TOKEN}",
    "Content-Type": "application/json"
}

payload = {
    "node_id": "slack-thread-12345",
    "prompt_template": "Executive Summary",
    "model_override": "gemma3-27b-it"
}

response = requests.post(API_URL, headers=headers, json=payload)
print(response.json())

Searching the Knowledge Base

# Search for specific documents via API
curl -X GET "https://your-workspace.curiosity.ai/api/v1/search?q=roadmap+2026" \
     -H "Authorization: Bearer <API_TOKEN>"
  • AnythingLLM — For flexible local RAG management.
  • Khoj — Open-source personal AI search.
  • Msty — Desktop-native local LLM interface.
  • Ollama — Primary local model provider for Curiosity.
  • Elasticsearch — For large-scale enterprise search infrastructure.
  • Authentik — For OIDC/SAML integration with Curiosity Workspace.
  • MCP Registry — For extending agentic context.

Sources / References

Contribution Metadata

  • Last reviewed: 2026-07-21
  • Confidence: high