AI that answers from your numbers.
The figures come from the engines: the model is handed values the app already computed and instructed never to invent one. The Strategy report checks every figure it prints against the engine; the chat is instructed the same way, but its replies are not checked figure by figure.
“What should I do with my Shopify position?”
It is rated HOLD.
↳ Source: Signals
The business has not deteriorated, so nothing here says sell — a SELL from this engine means the fundamentals broke, never that the price moved.
It is 14% above the blended fair value, which is rich rather than wrong.
↳ Source: Fair value
Your cost base is $8,410, so trimming would realise a gain you do not have to take this year.
↳ Source: Cost base
It sits in your TFSA, which is where it should be: no dividend to withhold.
↳ Source: This holding
An example answer. The model wrote the sentences; the engines produced every number in them.
The limits are the product.
In this category the live worry is that it will make something up about your money. So the claim that matters is the one about what it cannot do.
It is handed the numbers
Read-only tools hand it values the app already calculated, and it is instructed never to invent one. The Strategy report checks every figure it prints against the engine and flags a mismatch; chat replies are not checked figure by figure, so verify a number that matters.
Fair values, Deep Scores, your ACB, tax figures and the signal engine's output reach it through read-only tools.
It can draft. It cannot act.
Nothing is written to your portfolio until you press Confirm.
A confirmed draft takes the identical path as a trade you typed yourself — same ACB, same ledger.
It reads your numbers only
Answers are grounded in your holdings, your goals, your tax situation.
Candidate screens exclude what you already own, and multi-year decliners.
It tells you what it costs
You always see what a run will cost before it runs, and what the month has cost so far.
Checked on the server before every paid call, so the gate can't be bypassed from a browser.
Your technology exposure is 38% against the 25% single-sector limit you set. Two names are doing most of it, and one of them is also 11% of the whole portfolio on its own — above the 8–12% band its Deep Score earns, but not outside it.
Example answers. In the app these come back live, from your own holdings.
A macro-aware read on every holding you own.
Sample report on an example portfolio. Runs instantly and costs nothing.
In the app this reads your own holdings against the market regime — classified from five inputs: VIX, the US 10- and 2-year yields, the high-yield credit spread and the S&P 500 against its 200-day average — and returns a ranked plan. Press Generate to see the shape of one.
It is handed the numbers.
Read-only tools hand it values the app has already calculated — fair values, Deep Scores, your ACB, the signal engine's output — and it is instructed never to invent one. The Strategy report checks every figure it prints against the engine; chat replies are not checked figure by figure.
It can draft a trade. It cannot make one.
Nothing is written to your portfolio until you press Confirm, and a confirmed trade takes the identical path as a hand-entered one — same ACB, same ledger.
It reads your numbers, not everyone's.
Answers are grounded in your own holdings, your own goals, your own tax situation. Candidate screens exclude what you already own.
It tells you what it costs.
One budget meter across every AI feature, checked on the server before each paid call — so the gate can't be bypassed from a client.
Four jobs, each with a different question.
The grounded chat
Ask questions about your own money in plain English.
"Am I too concentrated?" "What should I do with this position?" "Can I hit my goal without more risk?"
Strategic Timing Advisor
Reads the economy today and tells you what it means for each holding you own.
Four market regimes, classified from five inputs — VIX, the US 10- and 2-year yields, the high-yield credit spread and the S&P 500 against its 200-day average — with rates, CAD/USD, oil and inflation shown beside them.
Tax & Structure Audit
Checks whether your accounts are set up right, rather than whether to trade.
Produces MIGRATE and CONSOLIDATE signals that no timing tool generates.
Sector Lab
Get properly good at one industry at a time.
A three-part thesis with clickable sources, then a private conviction journal for your own view and kill criteria.
The AI defers to the rules
Both AI tools are handed the rule engine's verdict on each holding and told to defer to it. A sell or trim either AI writes that the engine did not raise is shown as a point to review, never as a trade, and never stronger than the engine's own; so is a buy the Tax & Structure Audit writes.
A quality holding is never told to SELL on price alone. A SELL needs a low score the business itself confirms, dangerous leverage, shrinking revenue — or, in a contraction, a weak high-beta cyclical.
The trust boundary is visible
Web research and your own numbers are kept deliberately separate.
Five features read the web: Sector Lab, Earnings Recap, Winner Thesis, the Discovery search and the shared macro briefing (Canada CPI, headlines, sector notes). Each is marked as web-sourced where it appears, and Sector Lab's external claims carry a source you can click. Your own numbers never come from the web.
Nothing here runs without you saying yes.
Every paid action shows its cost first, checks your remaining budget on the server, and stops at the limit rather than surprising you at the end of the month. The report you just ran above cost nothing at all — it is a worked example, so you can see the shape of one before you ever spend a cent.