Early access

Yourmemory.

Never brief your AI twice. Dump anything — memax organizes, and every AI you use remembers.

MCPCLISDKWeb

Claude CodeClaudeCursorCodexGitHub CopilotWindsurfGemini CLIOpenClawHermesOpenCode

From connecting your agent to its first recall: under two minutes.

LongMemEval — the standard long-term memory benchmark

97.6%

retrieval recall@5

91.2%

QA accuracy · #9 worldwide

$0.046

per question, end-to-end

See the full benchmarks →

memax, everywhere you work

~/work/api — Claude Code

> why did we move access tokens to 1h expiry?

I'll check the team hub for that decision.

memax - recall (MCP)(query: "auth token expiry decision")

⎿ Found 3 memories (ctrl+r to expand)

April's security review flagged long-lived tokens — access tokens dropped to 1h, refresh stays 30 days. The full tradeoff analysis is in your team hub.

Claude Code pulls the answer from your memory, mid-session.

Claude Code — ~/finance

> save my ETF portfolio strategy to memax — the 70/20/10 split we just worked out

memax - push (MCP)(title: "ETF portfolio strategy")

⎿ Saved · 1 memory (ctrl+r to expand)

Saved to your personal hub. Any of your agents can recall it.

claude.ai

what's my latest investment portfolio strategy?

memax_recall

Your latest strategy (saved yesterday from Claude Code): 70% broad-market ETFs, 20% international, 10% bonds — rebalance quarterly, thresholds at ±5%.

One memory, every agent — saved from a coding session, recalled in chat. No platform's built-in memory crosses that wall.

Yourmemory.memory.

Yourmemory.