An independent lab

Durable, private memory for agents — so they're actually useful to humans.

Mindful Agents Lab builds long-lasting, privacy-preserving infrastructure for AI agents — memory that remembers, and storage that just works.

See what we're building

What we build

AI agents are only as useful as what they remember and where they keep it. Most forget everything the moment a session ends, and the storage they do have is brittle, opaque, or locked away.

Mindful Agents Lab builds the missing layer: durable, private memory that lets agents recall what matters over time, and agentic storage that works without friction. We think memory should last, stay private, and serve the person the agent works for — not the other way around.

What we're building

MEMORY

Opentusk

Memory for agents — the elephant that remembers.

A memory system that gives agents durable, searchable recall, so they carry context forward instead of starting from scratch.

Visit Opentusk ↗

STORAGE

Dockage.ai

Agentic storage on Walrus. S3 without the sign-up.

Object storage built for agents — decentralized, permissionless, and instantly usable. Store and serve data with no account and no friction.

Visit Dockage ↗

What the lab stands for

01 · ENDURING

Enduring

Built to persist. Memory and storage designed to outlast a single session, model, or vendor.

02 · PRIVATE

Private

Privacy-preserving by default. Your data stays yours, and stays yours to control.

03 · HUMAN

In service of people

Agents that are genuinely more useful — because the memory works for the human.

About

Mindful Agents Lab is an independent lab building foundational tools for AI agents at the intersection of memory, decentralized storage, and privacy. We keep the surface small, the defaults private, and the infrastructure built to last.