Structure, not signals.

The tooling gap in retail investing isn't data. Quotes, filings, transcripts and options chains are all free now, and a general-purpose model will summarize any of them on demand. What's missing is the layer a professional gets for free by working on a desk: someone asking, every week, which positions actually need a decision.

That's the gap Erina fills. It reads the book you already hold, surfaces the handful of decisions that are live this week, and reasons through each one in the open — then hands you two or three concrete options and gets out of the way.

What we won't build

Some of the more obvious things a product like this could do, we've deliberately left out. Each omission is load-bearing.

• Execution. Erina has no broker connection and no order path. Choosing a recommendation saves it for tracking; placing the trade stays with you. A tool that both recommends and executes collects an incentive we'd rather not have.
• Discretion. We're not a registered investment adviser and don't manage money. Nothing the product outputs is personalized advice, and every AI output says so inline.
• Signals without reasoning. There's no alert feed of bare buy and sell calls. If a recommendation can't carry an argument, it doesn't ship.
• Engagement mechanics. No streaks, no gamified trading, no nudges to act for the sake of activity. Holding is a first-class option in every workflow, and often the right one.

How we handle your positions

Your portfolio is the input that makes the product work, so it's worth being direct about it. Positions you upload are used to generate your analysis and nothing else — they aren't sold, and they aren't used to build a tradeable picture of what our users hold. Imports run through the same account you control, and you can delete a portfolio from the app whenever you like. The privacy policy has the full detail.

Where it's going

The six workspaces cover the decisions that come up most in a self-managed book. The direction of travel is depth rather than breadth — better reasoning inside each workspace, and better memory of what you decided last time, so the product gets more useful the longer your history with it runs.

If you're using it and something's wrong, we'd rather hear it early. The product is young enough that feedback still changes it.