Privacy Policy

Counsel to complete: the controlling legal entity, its address, and whether a Data Protection Officer or EU/UK representative is required must be stated here before publication.

• Account data — email address and authentication identifiers, handled by our authentication provider.
• Waitlist data — if you request early access, the email address you enter, held by our authentication provider so we can send you an invitation. Ask us and we will remove it.
• Newsletter data — if you subscribe to the weekly email, your email address, when you confirmed it, and when you unsubscribed. Nothing is sent until you confirm from the link we email you, every message has a one-click unsubscribe link, and erasing your account removes a subscription made while signed in. The address is passed to our email delivery provider to send each message.
• Alert email — if you switch on the daily alert email in Settings, each morning with new alerts we send them to your account's verified email address through our email delivery provider: the tickers and what changed, nothing from your analyses. It is off until you turn it on, and every message has a one-click unsubscribe link.
• Team data — if you join a team, the name you choose to be shown as, which the team's owner can see. The owner can also see how many analyses you ran on the team's plan and how accurate they were. Nobody on a team can see which companies another member analysed, or their portfolios.
• Usage data — the tickers you analyse, the analyses produced for you, chat messages you send to the consensus agent, the tickers and times of any batch you schedule, alerts about changes to the companies in your portfolios since your last analysis of them, any notes you add to your calls in the research journal, any thumbs up or down you give an answer and the reason you add to one, the names of any API keys you make (the keys themselves are stored only as a one-way hash), any thesis you write on a company with the conditions you set and how each report was checked against them, and counts of metered actions. A condition you write in your own words is checked by our AI provider's web search, which is sent that condition with the company's name and the report's figures - never your thesis statement. If you connect an AI assistant to your account (our MCP endpoint), with one of your API keys or by signing in and approving it, it can read the same data the API returns - your analyses, portfolios and calibration - and sends what it reads to the AI model it runs on, under that assistant's terms, not ours. We do not choose that provider or see what it does with the data. For an assistant you sign in, we keep its name, when you connected it and last used it, and its access tokens as one-way hashes; revoking the key or disconnecting the assistant in Settings cuts it off.
• Billing data — subscription status and customer identifiers. Card details are handled entirely by our payment processor and never reach our servers.
• Any credentials you choose to supply — if you provide your own model or data-provider API keys, they are encrypted at rest.
• Diagnostic data — error reports, which may include request metadata.
• Product analytics — a small set of events recording that you signed up, started or completed an analysis, downloaded a report, reached a plan limit, or clicked to upgrade, together with your account identifier and the pages you visit. These record what you did, not what you researched: the tickers you analyse, your own analysis text, and the content of any analysis or chat are never sent to our analytics provider. We do not record your screen or your keystrokes.

To operate your account, run the analyses you request, enforce plan limits, take payment, diagnose faults, and meet legal obligations. We do not sell personal data, and we do not use your private analyses or chat content to train models.

Counsel to complete: the lawful basis for each purpose (contract, legitimate interests, consent) needs to be stated explicitly for GDPR/UK GDPR.

The platform relies on sub-processors for authentication, hosting, database and cache infrastructure, payments, email delivery, error monitoring, product analytics, market and news data, and large language model inference. Content you submit for analysis is sent to the model provider configured for your account. Where an analysis includes web research, that provider also runs web searches for the company named and the report date; your own written analysis is not included in those searches. The tickers of holdings in uploaded portfolios are used to request daily closing prices from our market-data provider when a portfolio is uploaded and each night after, and options prices each night for holdings that report in the next few trading days; only the ticker is sent, with nothing that identifies you or your portfolio, and the prices saved are shared market data, not linked to your account. Once a prediction is scored, the saved debate and its outcome are sent to our model provider to write a short review of what went right or wrong; the review is stored with your account and shown only to you. When you ask for a written assessment of your results, the figures worked out from your scored predictions and their reviews are sent to the same provider; the assessment is stored with your account and shown only to you. When you ask for a written analysis of a portfolio, its holdings' tickers, weights, sectors and (on a plan that shows them) dollar values, their report dates, timings and implied moves, the lean and confidence of your own analyses of them, how many conditions of each thesis you hold on them held or broke at its latest check (never the thesis or its conditions in your words), how your stated confidence has compared with outcomes, and the Performance tab's backtest figures for the portfolio are sent to the same provider, and so are any follow-up questions you ask about it, with the same figures; the analysis and its questions are stored with your account, shown only to you, and deleted with the portfolio. Our product analytics are hosted in the United States.

Counsel to complete: a named sub-processor list with each one’s location and transfer safeguards. GDPR requires naming them, not describing them by category as above.

Counsel to complete: several likely sub-processors are US-based. If any user is in the UK or EU, the transfer mechanism (Standard Contractual Clauses, adequacy, or equivalent) must be identified here.

Depending on where you live, you may have the right to access, correct, delete, export or restrict processing of your personal data, and to object to it. You can request deletion of individual analyses or of your whole account at any time, and we action it manually.

Counsel to complete: the contact route for rights requests, the response deadline, and the relevant supervisory authority for complaints.

Access is authenticated on every request and scoped so that your analyses are readable only by your account. Any API keys you supply are encrypted at rest. Traffic is encrypted in transit. No system is perfectly secure, and we cannot guarantee absolute security.

Retention periods are set out separately in our Data Retention Statement.