Introduction
Atlas is a platform for AI-managed on-chain portfolios, built on Robinhood Chain. Its first product is a vault: an AI agent that manages a portfolio of tokenized real stocks and on-chain yield on your behalf. In plain terms: you give it some money, tell it how you want that money handled, and it takes care of the ongoing work of managing it — buying, selling, and shifting between stocks and yield — so you don't have to watch the market yourself. The same contracts and decision engine behind the vault are built to carry more than one product, which this documentation covers in full.
Why this needs explaining at all
If you've used a normal brokerage app, this might sound like nothing new. The part that's actually new is where your assets live. Robinhood recently launched its own blockchain and started issuing real stocks as tokens that exist on that blockchain — meaning a share of a stock (or something economically equivalent to it) can now be held in a crypto wallet, the same way you'd hold any other token, instead of being locked inside a brokerage account you can only access through one app.
That's a big deal, but it also creates a new question nobody has fully answered yet: if your stocks now live on-chain, who manages them day to day? A brokerage app doesn't manage your allocation for you — you do, or a human advisor does. Atlas's vault is built to be the automated version of that job, specifically for this new kind of on-chain stock ownership.
What you actually get
- A portfolio of Stock Tokens and yield-generating positions that stays aligned with rules you set, without you having to check in constantly.
- Every action taken on your behalf comes with a plain-English explanation — not just a transaction hash, an actual sentence explaining what happened and why.
- Full control at all times: you can see everything, change your rules, or pull your money out whenever you want.
The vault is the first product, not the only one
Everything described in this document — the vault, the manager role, the on-chain constraints — is a real, specific product: the first one built on Atlas's underlying engine. That same engine (the vault factory, the decision engine, the market signal pipeline) is designed to eventually support more than one product: infrastructure other teams can build their own vaults on, a signal feed other agents and researchers can query, and a marketplace for sharing allocation strategies. Where this documentation describes something specific to the vault, it says so; where it describes the platform underneath, it says that instead.
What this documentation covers
- The Portfolio Management Gap — why this problem exists and why nothing currently solves it.
- The Vault Model — where your money actually sits, and why the agent never touches it directly.
- How a Rebalance Works — the exact step-by-step of what happens when the agent decides to act.
- Constraints & the Trust Model — the actual limits placed on the agent, and how they're enforced.
- Use Cases — concrete examples of who this is for and what it looks like in practice.
- Supported Stock Tokens and Fee Structure — the practical details.
- The Platform — what's built on the same engine beyond the vault, and how it's sequenced.
- Security, Honesty & Roadmap — including what we haven't finished yet. We think this matters more than most documentation admits.
- FAQ & Glossary — quick answers and definitions if you just want the short version of something.
If you're new to any of this — tokenized stocks, on-chain vaults, or how an "AI agent" can safely execute real trades — keep reading in order. Nothing here assumes you're already a crypto expert.