Frequently Asked Questions
Common questions about Graphory, pricing, security, and how everything fits together.
Graphory is a graph-native data infrastructure platform. It connects to the business tools you already use (Gmail, Slack, QuickBooks, GitHub, and 20+ first-class connectors plus unlimited custom integrations), collects structured data from each one, and builds a unified knowledge graph. You query that graph through a REST API or by connecting your AI of choice via the Model Context Protocol (MCP).
Think of it as your business's memory layer - every email, invoice, calendar event, and code commit becomes a queryable, connected piece of your knowledge base.
Graphory is the memory layer - you bring any AI you want for analysis and synthesis. We provide tools via MCP that let your AI search the graph, explore relationships, view timelines, and write findings back.
Supported platforms include Claude Desktop, Claude Code, ChatGPT, Gemini CLI, Cursor, Windsurf, VS Code Copilot, Codex CLI, and any tool that can make REST API calls. See the Getting Started guide for setup instructions for each platform.
20+ first-class connectors ship with a maintained collector and a tuned extractor today:
- Communication: Gmail, Slack, Discord, Google Meet, Zoom
- Productivity and Project Management: Notion, Airtable, Linear, Jira, ClickUp, Basecamp, Todoist
- Developer Tools: GitHub
- Document and File Management: Google Drive
- Finance and Accounting: Stripe, QuickBooks
- CRM: Salesforce, HubSpot
- E-commerce: Shopify
- Scheduling and Calendar: Google Calendar
- Video and Media: YouTube
- Direct ingestion: File upload, generic webhook (Ingest API)
For anything else with a stable REST or GraphQL API, your AI defines a collector on the fly via
configure_custom_collector - we have walked Pipedrive, Outlook, Microsoft Teams,
Intercom, Zendesk, Dropbox, Snowflake, BigQuery, Asana, Calendly, and many more through this
flow in one MCP turn.
Two auth styles are supported: OAuth for major platforms (Google Workspace, Slack, Salesforce, etc.) and API key / personal access token for everything else (GitHub, Notion, Linear, Stripe, and the long tail). For every type, you supply your own credentials and Graphory encrypts and stores them per organization in your WorkOS Vault. You can paste them on the connections page yourself, or have your AI call our MCP tools to do it. When a credential expires or is revoked, we surface the provider's auth error back to you so you (or your AI) can rotate it. The only OAuth Graphory itself operates is WorkOS AuthKit for signup and login.
For anything without a collector, push data directly via the Ingest API. See the full list on the Data Sources page.
Yes. Each organization has its own isolated graph - no data is shared between organizations. Connections use OAuth, so you never share passwords with Graphory. You authorize access directly with each provider (Google, Slack, etc.), and you can revoke that access at any time.
API keys are scoped to a single organization and can be revoked immediately from the Settings page. We do not use your data to train any models.
Pricing is based on node count only. There are no limits on connections, entities, API calls, or team seats.
- Free: $0/month - up to 100K nodes
- Pro: $149/month - up to 500K nodes
- Business: $349/month - up to 2M nodes
- Enterprise: Custom pricing for larger deployments
Yes, Graphory is completely model-agnostic. It works with any AI that supports MCP or can make HTTP requests. This includes Claude (Desktop and Code), ChatGPT, Gemini, GPT-4, Llama, Mistral, or any other model. You can also call the REST API directly from Python, JavaScript, Go, or any language.
Generate an API key from Settings, then add Graphory's MCP server to your AI client's configuration file. The Getting Started page has ready-to-paste configurations for Claude Desktop, Claude Code, ChatGPT, Gemini CLI, Cursor, Windsurf, VS Code, and Codex CLI.
Yes, on Enterprise plans. Graphory's accuracy numbers on public benchmarks come from a general-purpose scoring pipeline. On Enterprise, we train a custom scorer on your labeled data - the same pipeline that added meaningful accuracy on Magellan in internal testing.
The output is a set of per-org weights that improve matching accuracy on your domain: your products, your customer naming conventions, your industry vocabulary. The tuned scorer lives inside your org and is used automatically for all entity resolution on your graph.
Contact support@graphory.io to discuss a custom tuning engagement.
Your data remains accessible for 90 days after cancellation under a soft-delete grace period. You can restore or export it at any time during that window via the API. After 90 days, your graph is hard-purged and cannot be recovered. OAuth tokens for connected sources are revoked immediately upon disconnection.