pumaDB
pumaDB provides a lightweight hosted memory layer for AI agents, enabling durable context preservation across sessions without complex database infrastructure.
At a Glance
- AI Agent Developers
- Users of LLM clients (ChatGPT, Claude, Codex)
- Serverless app developers
AI Tools by pumaDB
(1)
pumaDB
Hosted Memory Database for AI Agents
Discussions
No discussions yet
Be the first to start a discussion about pumaDB
Latest News
Products & Services
A hosted memory layer for AI agents that provides durable JSON storage via MCP (Model Context Protocol) or REST API.
Market Position
Positions itself as a simpler, faster alternative to vector databases and manual context management for small-to-medium agentic memory needs.
Leadership
Founders
Stuart Sim
Global Head of AI and Managing Director at Silver Lake. Formerly leader of Build by McKinsey (McKinsey & Company's community of technologists and designers). Background in building AI tools and platforms.
Executive Team
Stuart Sim
Founder
Managing Director and Head of AI at Silver Lake; former partner/leader at McKinsey & Company.
Founding Story
Created to solve the problem of context loss in AI agent workflows, where existing solutions were either too manual or too infrastructure-heavy.
Business Model
Revenue Model
SaaS / API-based (currently in early access/free tier with usage limits).
Pricing Tiers
20 tables, 1,000 rows per table, 25 MB total storage, 30 writes/min per key, 60 reads/min per key.
Target Markets
- AI Agent Developers
- Users of LLM clients (ChatGPT, Claude, Codex)
- Serverless app developers
- Reusable operating instructions for agents
- Project-specific workflows and repository facts
- User communication preferences and styles
- Research clippings and task state persistence
- Handoff context between sessions