Skip to main content

Aristo

Category: AI Infrastructure

AI agent memory layer startup building Membase, a persistent memory layer that unifies user context across AI agents and tools Aristo was founded in 2025. The company is led by Shinhyeok Hwang (황신혁). Based in Seoul, South Korea. Team size: 2-10. Latest round: Pre-Seed. Key investors include Kakao Ventures.

Founded
2025
Headquarters
Seoul, South Korea
Team size
2-10

Value proposition

Membase provides a unified, persistent memory layer for AI agents that syncs user context across all AI tools (ChatGPT, Claude, Cursor, Gemini, etc.), eliminating the need to repeat context when switching between AI services.

Products and solutions

Membase — a personal memory layer for AI agents that syncs context across ChatGPT, Claude, Cursor, Gemini, Codex, VS Code, and other MCP-compatible agents, integrates with Gmail, Slack, Calendar, and other work tools

Unique value

AI-service-agnostic personal memory layer that captures relationships between people, projects, decisions, and work history — not just raw data storage — enabling true cross-agent context sharing via MCP (Model Context Protocol).

Target customer

Individual developers, AI power users, researchers, and knowledge workers who use multiple AI tools and need persistent context; expanding to team collaboration and B2B memory API for AI agent developers

Industries served

AI Infrastructure, Developer Tools, Personal Productivity

Technology advantage

MCP-native architecture enabling cross-agent memory sharing; knowledge graph-based memory that captures relationships between entities (people, projects, decisions); automatic context sync from work tools (Gmail, Slack, Calendar); privacy-first with user-owned data, encrypted at rest and in transit

How they differentiate

Unlike agent-specific memory solutions (ChatGPT Memory, Claude Projects) that are locked to a single platform, Membase is AI-service-agnostic — it works across all major AI agents via MCP. Unlike general vector databases, it captures relationship context (people, projects, decisions) in a knowledge graph, not just raw embeddings.

Main competitors

Mem0 (open-source memory layer), Letta (formerly MemGPT, agent memory management), Zep (long-term memory for AI agents), Graphiti (knowledge graph memory by Zep)

Key partnerships

MCP (Model Context Protocol) ecosystem, compatible with Anthropic Claude, OpenAI ChatGPT, Cursor, Gemini, Codex, VS Code

Notable customers

Roy Tan (ex-Tech & Public Policy Head at Meta), Minsung Park (CEO of THEZONEBIO), Ignacio Montenegro (Entrepreneur), Collin Civish (AI Researcher) — early users cited on membase.so

Major milestones

2nd place at Anthropic Claude Code Builder Hackathon (Seoul, Nov 2025) judged by Anthropic co-founder Benjamin Mann, MVP built in 3 hours, Garry Tan (YC CEO) mentioned Membase on X, Public beta launch of Membase, Pre-Seed funding from Kakao Ventures (June 2026)

Market positioning

Early-stage startup in the emerging AI agent memory infrastructure space, targeting the growing pain point of context fragmentation across multiple AI tools. Competing in the "second brain" / persistent memory layer category alongside Mem0, Letta, and Zep.

Geographic focus

Global (North America primary focus for initial user acquisition; HQ in Seoul, South Korea)

About Shinhyeok Hwang (황신혁)

KAIST (Korea Advanced Institute of Science and Technology) graduate; previously at Kakao; AI research and product development background

Official website: