Getting started

This walks the whole product in the order it's meant to be used. Each step is genuinely useful on its own, so you can stop after step 2 and still be better off than you were. Budget about twenty minutes for the first three steps.

1. Sign in

Access is currently limited to invited accounts while the product is in its development phase. If you have one, sign in at the app.

Once you're in, the front door is Home — your book, what the AI thinks of it, and the trade plan that follows, threaded down one page. It'll be mostly empty until you finish step 2. The left rail groups everything into six sections: Home, Portfolio, Managers, Blend, AI, and System.

2. Get your holdings in

Everything downstream is only as good as this step, so it's the one to do properly. Open Portfolio → Add a source. Three ways in, and you can mix them:

• Sync a portfolio tracker. Connect your tracker account and your positions arrive on a schedule, with the sync's own health visible in-app. Best if you already use one.
• Import a CSV. Upload a broker export. You get a diff preview — exactly what will be added, changed and removed — before anything is saved.
• Enter positions by hand. Ticker, quantity, optional cost basis. Fine for a small book or a quick start.

All three land in the same place, and every number is stamped with how current it is.

What you have now: one view of everything you own across every account, with concentration made visible — your largest position, your top-three weight, and how many genuinely independent bets you're making once names that move together are counted as one bet. For a lot of people this alone is worth the trip.

3. Record the people you follow

Open Managers → Sources. A source is any reference portfolio that can feed your plan: a manager you track, your own written strategy, or an AI sleeve. Add one and record what they hold — with the date and where the information came from (a filing, a video, a member update).

Two things to set per source, because they're what make it useful later:

• Status — active, on trial, or just watching.
• How often they publish — so the app can tell you when their data has gone stale or overdue instead of letting you act on a book from March.

Recording a manager's book by hand is a chore, which is exactly why you don't have to: see step 6.

What you have now: a dated, attributed history of every book you follow. Record a second disclosure later and you can diff the two — including the exits nobody announced.

4. Blend them into a target

Open Blend → Workbench. This is the live, all-in-one view: drag a source's weight and the plan, the drift against your account, and the trade list all redraw at once. Nothing saves until you say so, so experiment freely.

• Set a trust weight per source — how much of your plan that person is responsible for. Weights normalize to 100%.
• Set the rules you always want applied — names you'd never hold and where their money goes instead, a cap on any single position, how much of the portfolio stays in cash. The full editor is on the Blend designer page, and every save is versioned.
• The result is your target portfolio. Open any line to see which sources contributed it and at what weight. Excluded names are listed too, with your reasons.

What you have now: an explainable plan. Not "the algorithm says 9.5%" — these two managers hold it, at these weights, under these rules you set.

5. See the gap, and decide

Open Blend → Reconcile, pick a saved target and one account.

You get the position-by-position drift and the trade plan that closes it: the buys and sells, sized, with the cash implication. Export it to CSV or print it and run it at your broker.

It's a proposal. Nothing here places an order.

While you're on this page, spend two minutes on conviction custodians: anchor each position to the reason you hold it — a source, your own thesis, or the system. Right after an import everything is unanchored, which is normal. Once anchored, the app can tell you when a reason disappears — a manager exits a name and you're suddenly holding an orphan.

6. Work by talking instead of clicking

Anchoring 80 positions or transcribing a manager's book is the kind of machinery this product is supposed to spare you. So connect your AI assistant — Claude or any assistant that speaks MCP — to your account.

From your normal chat window it can read your holdings and sources, record a manager's new disclosure, propose custodian anchors for your whole book with reasoning, and flag what's drifted. You review and approve in conversation; it does the typing.

It cannot trade. The tools to place an order, move money, or touch a brokerage account do not exist on that surface — not as a rule it's asked to follow, but as tools that aren't there to call.

7. Ask the panel (Pro)

The AI Council is a set of AI portfolio managers with different investing philosophies, grounded in your actual holdings. Talk to one, or convene the panel to debate a question and return prioritized recommendations with reasons.

Available on the Pro and Ultimate tiers.

8. Keep score (optional, and worth it)

Open Managers → Predictions. Record what a manager actually claimed — verbatim — and what would have to be true for it to count as right. Later, you rule on the outcome and their scorecard builds up per kind of claim.

It's the honest input to the trust weight you set back in step 4.

9. Automated execution (Ultimate, optional)

The top tier adds an automated execution module. It starts on a paper (simulated-money) account, can never open an undefined-risk position, and needs two separate deliberate actions from you before it can touch real money. Read that page before connecting anything.