Stoke Money Forecasts Daily Cash Flow Without a Budget
Stoke Money builds a day-by-day forecast for each account from repeating transactions in your history, showing what is safe to spend without keeping a budget. Users can link a bank account or enter accounts and transactions by hand, and both feed the same forecast. Stoke Money
Detection starts with history. The system spots paychecks, rent, subscriptions and bills, then groups them into repeating obligations. From those patterns it lists when bills land, what the buffer looks like after they clear, and whether rent will clear on its due date. No budget categories are required. The question shifts from how spending is labeled to whether cash will be present on a given day.
The spending record and the forecast use the same underlying data, reconciled daily. Cash is treated as leaving when the card is paid, not at each swipe. A card purchase appears in spending on the swipe date, while the forecast projects each card payment on its due date from the paying account. That split separates attribution from liquidity. Spending analytics stay tied to transaction time. The cash projection stays tied to when money actually moves.
Ingestion is either connected or manual. Bank sync runs via Plaid, a third-party connector that provides read-only access to transaction data. Stoke Money states it never sees or stores bank logins and cannot move money. Plaid coverage reaches most banks in the United States and Canada but not all, and unsupported accounts can be hand-entered into the same forecast. Manual entries are not a side ledger. They flow through identical projection logic.
Bank access tokens and account numbers are encrypted with AES-256-GCM, a widely used strong encryption standard, before storage. Data travels over TLS, the standard protection for data in transit on the web. Sign-in is via passkey, a device-based login that replaces passwords, or one-time email link, with no passwords to phish or reuse. The company states it makes money from subscriptions and never sells data or uses it for advertising. Stoke Money
For AI access, Stoke Money does not include a chatbot. Pro adds an MCP connection, a standard link that lets an outside AI app read data, from the user's own AI app to Stoke data, with support listed for Claude, ChatGPT or another MCP-capable assistant. That keeps inference and prompting outside Stoke's boundary. The model context comes from the user's chosen client, with Stoke acting as a scoped data source rather than a conversational host.
Pricing is subscription only. Plus is listed at $7 per month or $60 billed yearly. A $12 per month or $96 billed yearly tier includes connecting an MCP-capable assistant, which the site associates with Pro. Cancelling moves the user to the free plan rather than deleting the workspace outright. Stoke Money
The broader context here is the long failure mode of personal finance software. Transaction categorization got very good. Forward visibility did not. Most tools explain last month with precision and leave next Tuesday vague, especially when credit cards decouple purchase dates from cash outflow dates. Unifying those two timelines in one projection addresses the actual overdraft risk, which is almost always a timing problem rather than a categorization problem.
In my view, the architecture choice worth watching is MCP exposure without a native chatbot. It reduces Stoke's prompt-injection surface, where hidden instructions in data try to steer an AI, and avoids storing chat history alongside financial tokens, but it pushes access control, retention and leakage risk to whatever client the user connects. For tech-literate users that tradeoff makes sense. Scoped, read-only financial context served to a model you already trust is more flexible than another embedded assistant, and it leaves open what that assistant can compute once it can see daily balances, due dates and buffers.
