HaystackedAI

dividende-ai.com is a dividend calendar built for trust. Income investors need to know what each holding will pay, when it will pay, and whether that payout is safe — but free tools mostly show past dividends and unreliable estimates, rarely say what is confirmed versus guessed, and usually surface cuts only after the board announces them. Coverage of Canadian (TSX) stocks is especially thin. dividend-ai.com fixes this by living where investors already look: the calendar. Subscribe once in Google, Apple, or Outlook and upcoming payouts appear alongside everything else, with each amount clearly labeled as confirmed or projected. The key design rule is that facts never come from an AI model. Declared amounts are taken deterministically from filings and dividend data sources. The AI only does the 'leading read': it estimates the next payout from payment patterns, flags dividends at risk of a cut or raise, and explains its reasoning with sources. Every row carries one of three labeled layers — fact, pattern estimate, or AI prediction — each with its own confidence, reasoning, and sources. When a company actually declares, the prediction automatically corrects itself to the declared amount, so the calendar gets more accurate over time and builds a verifiable track record. Beyond the calendar, dividend-ai.com is a personal income tracker: enter your trades and see what your holdings will pay you and when, including special dividends, cuts, and variable payers that simple yield screens miss. An hourly market briefing keeps the calendar fresh and gives investors a reason to return. It is built for self-directed dividend and income investors in Canada and the US first — including FIRE savers and retirees living off dividends — with TSX tickers, CAD amounts, and North American time zones handled natively. What makes dividend-ai.com different is process, not just a model: deterministic truth grounding, labeled confidence on every number, calendar-native delivery instead of another dashboard to check, and a self-correcting record that lets users verify accuracy for themselves.