Area
pumaDB
Notion
Primary job
A small hosted JSON database for agent memory and lightweight server-side app state.
A collaborative workspace for team docs, wikis, projects, databases, AI tools, and automations.
Data model
Named JSON tables. Current account limits are 20 tables, 1,000 rows per table, 25 MB per account, and 64 KB per row.
Pages and blocks organized into workspaces, databases, properties, relations, views, templates, docs, wikis, and projects.
Query behavior
CRUD, simple equality filters, sorting, counting, batch writes, and short-lived viewer links for larger results.
Workspace search, database filters and sorts, linked databases, views, relations, rollups, and AI-assisted answers across workspace content.
Agent interface
Hosted Streamable HTTP MCP, local stdio MCP, REST API, and CLI are built into the product surface.
Agents can use Notion through official APIs, connectors, automations, or custom tooling, but the core product is a human collaborative workspace.
Collaboration model
Designed for agents and server-side apps to share small, explicit records across sessions.
Designed for people and teams to collaborate in shared pages, databases, permissions, comments, mentions, and workspace workflows.
Shape of memory
Structured rows are easy for agents and apps to validate, update, filter, restore, and treat as durable state.
Flexible pages, blocks, and databases are easy for teams to read, edit, organize, present, discuss, and reuse.
AI layer
Stores memory as explicit JSON records. It does not run workspace search, meeting notes, automations, or AI agents.
Positions Notion AI around workspace search, custom agents, meeting notes, writing help, app connections, and automating busywork.
Recovery model
Every update and delete archives the previous row. The last 10 versions are kept for 30 days and can be restored.
Relies on workspace/page history, permissions, trash, export, and plan-specific admin or enterprise controls.